1 Wednesday, 27 September 2023 2 (10.00 am) 3 ANNE OLIVIA CHAMBERS (continued) 4 Questioned by MR BEER (continued) 5 MR BEER: Good morning, sir, can you see and hear 6 me? 7 SIR WYN WILLIAMS: Yes, I can. 8 MR BEER: Apologies for the delay, we had a very 9 thorough fire drill this morning. 10 Good morning, Mrs Chambers. 11 A. Good morning. 12 Q. Can we start, please, with the piece of homework 13 that I think we set ourselves and, just to 14 remind ourselves of some context, can we start, 15 please, with LCAS0000224. This was the document 16 we were looking at, if you remember? 17 A. Yes. 18 Q. You told us this was an extract, I think, of 19 some ARQ data? 20 A. It appeared to be, yes. 21 Q. You noted that the selection that had been made 22 at the top under "Keys" required this to be in 23 date and time -- chronological order and the 24 data wasn't -- 25 A. Yes. 1 1 Q. -- and that it wasn't a complete record, because 2 there were lots of things that weren't on it? 3 A. Yes. 4 Q. The first thing to say is that we've researched 5 the document. The document was also provided to 6 us by Fujitsu in printed copy and in Excel 7 format and the Excel document is in exactly the 8 same format as this. So if any manipulation -- 9 I think you used that word neutrally -- had been 10 done, it had been done before it left Fujitsu? 11 A. Can I just say, though, that we do have 12 a document which has got all this data in. 13 Q. Yes, I am going to move to that in a moment. 14 A. Okay. 15 Q. It was just to clear up the point, in case 16 anyone thought that, because this was produced 17 with LCAS, a Lee Castleton number, putting two 18 and two together with what you said about 19 manipulation, anything had gone on there. We 20 have exactly the same document, in printed and 21 in Excel format, from Fujitsu? 22 A. That certainly was not my intention -- 23 Q. I know but I need to clear these things up. 24 Can we look then, please, at LCAS0001322. 25 If that can just be expanded so we can see the 2 1 whole of the page, please. Thank you. That 2 makes it not particularly legible but we can see 3 what it is and, if we just go to the next page, 4 please, we can see that that data, I think, ends 5 on 8 January? 6 So if we go back to the first page, please, 7 and expand, please, so we can see better. Thank 8 you. Is this also ARQ data? 9 A. Yes. 10 Q. Does this appear to be a more complete record of 11 ARQ data -- 12 A. Yes. 13 Q. -- and does this appear to be ordered in 14 ascending date and time order? 15 A. Yes. 16 Q. Okay. Can we scroll down, please. Can you see 17 an entry about 12 lines -- I'm not sure we're on 18 the right page. 19 A. Is it a little further up? 20 Q. It's the first page, please. We're on the 21 second page at the moment. So the first page, 22 please. Thank you. Then scroll down, thank 23 you, and scroll down some more, please. 24 Can you see an entry -- 25 A. Yes, I can. 3 1 Q. "Declare stamp total 118,322", yes? 2 A. Yes. 3 Q. That's the same as on the extract, as I'm going 4 to call it, that we looked at yesterday -- 5 A. Yes. 6 Q. -- but different from the figure in the cash 7 account total? 8 A. Yes. 9 Q. That seems to have been declared at 17.06.59; 10 can you see that? 11 A. Yes, yes. 12 Q. Then is there anything else relevant on that 13 page? I'm not sure that there is. 14 A. No. 15 Q. Then go over the page to the next page, please. 16 Can you see about 12 lines in, "ONCH total, 17 43,013.92" -- 18 A. Yes. 19 Q. -- at 17.31.20? 20 A. Yes. 21 Q. What does that mean? 22 A. That's the Overnight Cash Holding Declaration, 23 so they should, at that point, have counted what 24 cash they'd got and declared it for the -- as 25 an ONCH declaration. That was something they 4 1 had to do at the end of each day. 2 Q. Yes. Can you see about three lines in to the 3 second piece of highlighting "Declare cash total 4 £43,013.92", so the same figure, yes? 5 A. Yes, I can. 6 Q. It says, "Declaration complete". What does that 7 mean? 8 A. They've done a proper cash declaration at that 9 point. That was something they only had to do 10 once a week, as part of the balancing process 11 but you'd expect it to be the same figure as the 12 Overnight Cash Holding for that day. 13 Q. And it is? 14 A. Yes. 15 Q. We then see "Report cash declared" at 17.38.43, 16 "Report printed". So is that them in the branch 17 printing off a copy of the cash report? 18 A. That's the printing off of a copy of the cash 19 declaration report. So that will show what 20 they've declared by denomination of cash because 21 that's how it was done. 22 Q. Then next entry "Declare cash total", it's gone 23 up by about £300; can you see that? 24 A. I can see that. 25 Q. Then that is printed -- 5 1 A. Yes. 2 Q. -- a couple of seconds later? 3 A. Yes. 4 Q. Then "Declare stamp zero"? 5 A. Yes, I can see that. 6 Q. I couldn't see, on here, a record of the stamps 7 held in stock of the £5,500-odd? 8 A. No, and I can now explain why that was. 9 Q. Yes, please. 10 A. Okay. When the postmasters declared stamps, 11 this was just the stamps that were 1p, 2p, 12 that -- stamps that have a specific denomination 13 on them. So they would be presented with 14 several screens and they go through and say how 15 many of each denomination they've got, so that 16 is their stamp declaration. 17 Other types of postage, including First and 18 Second Class stamps and stamp books, and some 19 other things, were not included in that 20 declaration. They would be managed either by 21 "declare stock" or by "adjust stock", which is 22 what Mr Castleton did. So this stamps -- 23 postage stamps total that we've got is only for 24 a subset of the overall postage. 25 We can see that it's not at all clear on the 6 1 cash account. I think I say in my witness 2 statement that the cash account was a horrible 3 document to try to understand, partly for this 4 reason, but if you look at the final balance 5 report, which precedes the cash account and is 6 using the same data that the cash account is 7 built from, you can see on that final balance 8 report the section for postage, which includes 9 both the postage stamps as declared and all the 10 other types of postage. And if you want to look 11 at that, I can give you a reference. 12 Q. It's okay, I've got that. 13 A. You've got that. 14 Q. So, in short, this more expanded set of data 15 doesn't tell you anything that we didn't know 16 last night? 17 A. No. 18 Q. You've just remembered something? 19 A. Just by looking at that final balance report, it 20 reminded me. 21 Q. Thank you. Yes, that can come down. 22 Can we turn up -- I'm coming to the closing 23 of the PEAK now. Can we have up, please, 24 FUJ00146165. Thank you. This is the PEAK that 25 we were looking at yesterday in relation to your 7 1 investigation and can we turn to page 2, please, 2 and look at the foot of the page. Can we look 3 at your entry of 15.16.54. This is essentially 4 your closing entry on the PEAK; is that right? 5 A. Yes, it is. 6 Q. I don't think we've read this but we ought to 7 read this because it forms the substance of the 8 evidence that you give in your witness 9 statement, both in the civil proceedings in 2006 10 and in your witness statement to this Inquiry. 11 You say: 12 "Checks are ok. Cheques are being handled 13 correctly (except for [the 10th] when the clerk 14 forgot to cut off the report -- but this didn't 15 cause a discrepancy). Cash declarations look 16 ok, they usually use drawer ID 11. Occasionally 17 they have a different drawer ID, this can lead 18 to amounts apparently doubling on the cash flow 19 report, and should be avoided." 20 Can you explain what you meant by that, 21 please? 22 A. If they declare the same cash with two different 23 drawer IDs then when the Overnight Cash Holding 24 figure is sent off to the central systems, that 25 will add those amounts together. So if they've 8 1 declared it with drawer ID 11 and 22, for 2 example, and in both cases £10,000, then the 3 total for the day will appear to be £20,000, and 4 that's what will be sent to the cash handling 5 centre but those figures are not used in the 6 balancing process. 7 Q. That's why you said, "but again it will not 8 cause a discrepancy"? 9 A. Yeah. 10 Q. Why will it not cause a discrepancy? 11 A. Because these figures aren't input to the 12 balancing process. 13 Q. You continue: 14 "Checking the cash transactions on the 15 system against the declarations shows they are 16 not working particularly accurately ..." 17 The "they" in that sentence, who is that? 18 A. The branch. 19 Q. "... (ie at the end of the day the cash they 20 declare in the drawer is tens, hundreds or 21 thousands of pounds astray from what has been 22 recorded on the system)." 23 A. Yes. 24 Q. "It is possible" -- 25 So how did you establish that? 9 1 A. I established that by pulling out all the cash 2 transactions for a period -- and I now cannot 3 remember the precise length of that period -- by 4 taking the opening figure of cash, which you can 5 cross-check on the balance reports, and so on, 6 so you put that at the top of your column in 7 an Excel spreadsheet and then you work out the 8 running total by making all the cash adjustments 9 from that base figure, so at any point you can 10 see what cash the system thinks the branch 11 should hold. 12 Then from the Overnight Cash Declarations 13 and also the declarations made when they 14 balanced, you can see what the branch is 15 declaring that they hold and then it's simple 16 arithmetic to look for the difference between 17 those figures. 18 Q. You say: 19 "It is possible that they are not accurately 20 recording all transactions on the system. There 21 is no evidence whatsoever of any system problem. 22 I've mentioned this outlet to Julie Welsh 23 (Customer Services) who will try to get [the 24 Post Office] to follow it up, but in the 25 meantime please tell the [postmaster] that we 10 1 have investigated and that the discrepancies are 2 caused by the difference between the 3 transactions they have recorded on the system 4 and the cash they have declared, and are not 5 being caused by the software or hardware." 6 Then just over the page, please: 7 "I have checked various things on the 8 system. All the internal" -- 9 I'm so sorry, I think that's a repetition of 10 a message, isn't it? 11 A. Yes, I'm not sure why that's there. 12 Q. That's essentially the first sentence of your 13 15.16 entry repeated. 14 A. Yeah. 15 Q. In any event, you then essentially close the 16 call -- 17 A. Yes. 18 Q. -- and give it a category number, "Advice and 19 guidance given"? 20 A. Yes. 21 Q. Is that advice and guidance given back to the 22 HSH? 23 A. That was just the closure category that I chose 24 to use. 25 Q. Yes, but what does it mean? 11 1 A. It just means that I felt, out of all the 2 closure categories open to me, that was the best 3 one that meant I had given advice, that I had 4 written down, as in this case, some advice and 5 guidance to be passed back to the postmaster. 6 Q. Through the HSH? 7 A. That would be the normal route but, although -- 8 I had said in my response, I was also passing it 9 back to Post Office via Julie Welsh. 10 Q. Time -- it says: 11 "Hours spent since call received: 0 hours." 12 Why does it record that? 13 A. Because that wasn't a field that we ever used. 14 Q. Okay, you didn't record how long it took you to 15 do things? 16 A. No. 17 Q. Okay. Then the call was closed? 18 A. Yes. 19 Q. Can we turn up your witness statement, please, 20 at WITN00170200, and look at page 14, please. 21 Can we read together paragraphs 41 and 42 to 22 start with. You say: 23 "I felt that the only way to progress the 24 investigation was at the branch, to check they 25 were following the correct business processes 12 1 and correctly recording everything that was 2 done. Fujitsu did not have staff who were 3 allowed to go to branches to investigate 4 business or financial problems; this was Post 5 Office's responsibility. 6 "At this point, having not found any 7 evidence of a system error, the correct process 8 was for me to close [that's the PEAK we've just 9 looked at] with my findings, and HSH would 10 communicate them to the [subpostmaster]. 11 However, given the number of calls that 12 Mr Castleton had already raised, I felt this was 13 not in his best interests, as the problem 14 persisted, and so I contacted Julie Welsh in 15 Customer Services, knowing that she had a route 16 to the Post Office and hoping that this would 17 help Mr Castleton to find the cause of the 18 losses. Then I closed [the PEAK] 'Advice and 19 guidance given', stating what I had found and 20 that I had contacted Julie Welsh." 21 Then if we turn to paragraph 51 of your 22 witness statement on page 17, please -- 51, 23 thank you -- you say: 24 "I did consider the issues that Mr Castleton 25 was reporting to be unusual, and that he needed 13 1 assistance to get to the bottom of it. I was 2 not in a position to give or organise that 3 assistance." 4 A. Yes. 5 Q. The way that you set matters out in those three 6 paragraphs, 41, 42 and 51, may suggest that you 7 didn't think that, because you couldn't find 8 a system error, Mr Castleton or his staff must 9 have either taken the money or have been 10 responsible for the missing money. There was 11 a business or a process issue that still 12 required to be investigated? 13 A. Yes. I mean, it wasn't for me to decide what 14 was wrong but, from what I could see, that -- 15 the way the figures were swinging around so much 16 each day, that was really unusual, and so -- and 17 I could see from the calls that Mr Castleton was 18 raising that he was desperate for somebody to 19 help, and I wanted to try to, you know, progress 20 it -- I certainly didn't just want to send the 21 call back saying, "Oh, not us". I wanted to see 22 if we could get him some more help, maybe so 23 that he could do -- to help him sort of keep -- 24 try and keep a manual record or for somebody to 25 monitor what was happening. 14 1 I didn't know what the options were as 2 I said, this -- Fujitsu hadn't -- had no way of 3 helping with anything like that at branches. 4 Q. So why did you think it wouldn't be in 5 Mr Castleton's best interests just to close the 6 call and let the Helpdesk report your findings 7 back to him? 8 A. Because then it would just have gone back 9 saying, "Not a system error, talk to NBSC", and 10 NBSC would have said, "Oh, well, we can't find 11 anything, talk to the Helpdesk", and -- 12 Q. It would have gone back into the big machine? 13 A. Yes, he had already obviously spent an awful lot 14 of time on the phone trying to get somebody to 15 take some notice. 16 Q. What did you expect Julie Welsh in Customer 17 Services to do? 18 A. I knew that she had somebody in Post Office that 19 she talked to about -- well, I mean, it was part 20 of her role, I think, to coordinate with Post 21 Office, although I wasn't entirely sure on what 22 level. She was one of these helpful sort of 23 people, so I asked her if she could pass the 24 information on to Post Office. 25 Q. So she's in Fujitsu's Customer Services -- 15 1 A. Yes. 2 Q. -- but she had a line in to the Post Office -- 3 A. Yeah. 4 Q. -- almost outside of the Helpdesk and NBSC 5 machine? 6 A. Yes. 7 Q. So you didn't know exactly what she might do 8 but, being a helpful person, you thought she 9 might have some influence or connections? 10 A. I thought she would probably pass my findings on 11 to somebody within Post Office who would then be 12 able to take further action as they thought 13 appropriate. 14 Q. You'd said in the entry on your PEAK -- I'm not 15 going to go back to it -- but you passed it to 16 her "to try to get POL to follow it up". What 17 could POL do to follow it up, in your mind? 18 A. Go to the branch and, as I said, maybe help him 19 try and run a manual process alongside to 20 record, so that you could, at the end of the 21 day, look for any anomalies, you know, if there 22 was something that they were doing wrong 23 process-wise, to find that. 24 You know, at this stage, I sort of assumed 25 that Post Office did have people who were -- who 16 1 would do that sort of role. 2 Q. Were you referring it to Julie Welsh in order 3 that the next step could be an audit accompanied 4 by an immediate suspension on the day? 5 A. That was certainly not in my mind, no. 6 Q. That can come down, thank you. 7 Can we move forwards, please, to you giving 8 evidence and you being selected as a witness. 9 Can we start, please, by looking at POL00090437, 10 page 69, please. Thank you. 11 We can see that this is a letter from Rowe 12 Cohen Solicitors, dated 25 July 2006 to the Post 13 Office's solicitors, Bond Pearce, about the case 14 of Post Office v Mr Castleton. 15 A. Yes. 16 Q. It's a two-page letter, yes? 17 A. Yes. 18 Q. I think you were asked to review this; is that 19 right? 20 A. Yes, I believe so. 21 Q. If we scroll through it and you just read it to 22 yourself as we scroll through it, to refresh 23 your memory. If you look, in particular, at the 24 second, third, fourth and fifth paragraphs. 25 A. Yes, I certainly saw this at the time, whether 17 1 I -- and then we did some work as a result of 2 this letter, which isn't -- yeah. 3 Q. Can you remember who asked you to look at it? 4 A. I think probably this came through Gareth 5 Jenkins but it's possible that Brian Pinder or 6 somebody else would have showed it to me, but 7 I really I can't remember. 8 Q. Can we look, please, at -- so, essentially, I'm 9 going to summarise it. It was said by 10 Mr Castleton's solicitor that some analyses had 11 been conducted of some weeks, week 42 in 12 particular, and Mr Castleton believed there were 13 some incomplete transactions and other 14 discrepancies, yes? 15 A. Yes, yes. 16 Q. Can we look, please, at FUJ00152295. Can we 17 see, please the title at the top "Initial 18 Comments on Marine Drive letter", yes? 19 A. Yes. 20 Q. I think that's a reference to the letter we've 21 just looked at? 22 A. Yes. 23 Q. We can see this is dated 17 August 2006? 24 A. Yes. 25 Q. Can you see that? Then if we go to the second 18 1 page, please, and to the bottom, we can see that 2 it's your document? 3 A. Yes, that's right. 4 Q. Yes? 5 A. Yeah, yeah. 6 Q. Then back to the first page, please. You go 7 through parts of the letter, yes? 8 A. Yes, I do, yes. 9 Q. If we look at the fourth paragraph on the page 10 there, you say: 11 "Looking in the full message store at the 12 first of these ..." 13 A. Yeah. 14 Q. What are you referring to there "the first of 15 these"? 16 A. By "the full message store", at this point 17 I mean the archived messages -- sorry, audited 18 messages for Marine Drive over this period. So 19 at the time, back in 2004, all messages produced 20 at Marine Drive and everywhere else would have 21 been sent off into the audit files, from which 22 subsequently the ARQ data could be extracted. 23 But it was also possible just to look at the 24 full set of messages, rather than the specific 25 messages in -- that were -- that formed the ARQ 19 1 extracts. 2 Q. Thank you. You say: 3 "... I found the missing transaction was 4 a stamp sale within a Smartpost session, and the 5 Start information had not been included in the 6 message." 7 A. Yes. 8 Q. Can you explain what you mean by that, please? 9 A. Yes, because there were two sessions that didn't 10 net to zero in the ARQ data -- 11 Q. You wanted to know why? 12 A. I wanted to know why, yes, and it turned out to 13 be that there was a line, a transaction line in 14 the full message store that had not been 15 included in the ARQ extract. 16 Q. Is that a bug? 17 A. It was a bug in the ARQ -- well, it was a bug 18 whose only consequence was in the ARQ extract. 19 Q. You say: 20 "This was a not uncommon problem at the 21 time." 22 A. It was something that did show up in the 23 reconciliation reports, which I hadn't seen the 24 January ones because they no longer existed when 25 I'd looked originally. And I can't quite 20 1 remember but perhaps every few days you might 2 have one or two entries in the reconciliation 3 reports. This was for all 18,000 branches, so 4 it wasn't that it was happening every day for 5 every branch, but it was a problem that we knew 6 about and had been looked at, and the only 7 problem was that these particular smart mails, 8 Smartpost transactions were slightly malformed, 9 in that the start date attribute was missing, 10 which should have been they're. 11 Now, that didn't affect balancing in the 12 least but, because the ARQ extract extracted on 13 start date and it didn't have a start date, it 14 didn't find these. But they were included in 15 the branch balance. 16 Q. Which is what you say next: 17 "The transaction was included in all the 18 accounts (otherwise there would have been 19 a receipts and payments mismatch), and did not 20 cause any discrepancy at the branch but has not 21 been retrieved from the archive by the data 22 retrieval, since that requires Start date to be 23 present (see the query at the beginning of the 24 transaction log spreadsheet)." 25 Then you say, "Possible further action", if 21 1 we just scroll down: 2 "(a) confirm second instance is the same (so 3 far only have CAP42 complete messages)." 4 I don't understand what that means. Can you 5 explain please? 6 A. The whole message store extract, rather than 7 just the ARQ data. At the point I wrote this, 8 I'd only got the complete set for CAP42, for 9 week 42. 10 Q. I see: 11 "retrieve the TPSC253 report for the days 12 (Counter Exceptions), which should show the 13 problem was noticed at the time. MSU and SSC no 14 longer have this report but maybe we can get it 15 back through the Audit team." 16 A. Mm. 17 Q. Can you explain what you were suggesting there? 18 A. This was the reconciliation report I was talking 19 about, that we only kept for a very short length 20 of time. That's suggesting that maybe it was 21 a file that was audited that we could have 22 retrieved. As I've said, SSC knew nothing, 23 really, about audit and didn't know which files 24 were retained. I'm obviously wondering there 25 if, actually, it's something that they did have 22 1 that I didn't know about. 2 Q. Okay, and then (c): 3 "confirm in some way that no complete 4 sessions are missing for the same reason." 5 Yes? 6 A. Yeah. 7 Q. So, to summarise, there was a small discrepancy 8 caused by a missing Smartpost stamp sale, yes? 9 A. It's not a discrepancy. 10 Q. Sorry -- 11 A. A difference. 12 Q. -- a difference, thank you, on the data. That 13 wasn't an uncommon problem at the time but there 14 would be a daily report which would or should 15 confirm whether or not that difference was 16 picked up at the time -- 17 A. Yes. 18 Q. -- and that there might be a way formally to 19 check whether any complete transactions were 20 missing as a result of the known fault? 21 A. Yes. 22 Q. Okay. So they were your suggestions for further 23 action or possible further action? 24 A. Yeah. 25 Q. Can we look, please, at WBON0000027. Thank you. 23 1 Your document was turned into a report written 2 by, I think, you and Mr Jenkins; is that right? 3 A. Yes, he took responsibility for it. Yes. 4 Q. But this is jointly authored; is that right? 5 A. He authored it but it is based on stuff that I'd 6 investigated. 7 Q. Okay, and this is it, yes? 8 A. Yes. 9 Q. If we can read the first part of paragraph 1: 10 "As part of a prosecution associated with 11 Marine Drive Post Office ..." 12 Did you understand that Mr Castleton was 13 being prosecuted? 14 A. Yes, at this point, because I'd already had to 15 produce a witness statement. 16 Q. What would you understand by prosecuting? 17 A. That he was being taken to court, I suppose. 18 Yes, my legal knowledge is not very good. 19 Perhaps it's the wrong word. I don't know, 20 I didn't write -- I have no idea if that's right 21 or not. 22 Q. In your mind, was there a distinction between 23 a criminal court and a civil court? 24 A. Yeah, I think I knew there was a difference and 25 I knew that he was not being charged with any 24 1 criminal offence but Post Office were wanting to 2 get the money that they believed was owed to 3 them and that it was a civil matter. 4 Q. "... Anne Chambers and Gareth Jenkins have 5 undertaken an analysis of all transactions that 6 took place during Cash Account week 42 in that 7 Office." 8 A. Yes. 9 Q. Those transactions are between dates and times 10 set out, yes? 11 A. Yes. 12 Q. "The purpose of this note is to outline what was 13 done and also comment on the letter sent on from 14 the Post Office solicitors where Mr Castleton is 15 claiming to have shown that the Horizon figures 16 are incorrect." 17 Then scroll down to "Analysis undertaken": 18 "The initial set of data obtained was the 19 extract from the Transaction Log that was 20 submitted to Post Office as supporting evidence 21 (ARQ 421)." 22 What does the number refer to? 23 A. I think each ARQ request had a unique number. 24 Q. I see: 25 "Subsequently a complete extract of audit 25 1 data for the period concerned was obtained. 2 This included non-transactional data (including 3 opening figures) and the electronic Cash Account 4 information (which was subsequently submitted to 5 Post Office Limited's back end systems) and 6 represents the same information as was printed 7 on the paper cash account which Mr Castleton 8 signed at the time to indicate that it was 9 correct. 10 "The figures examined have been [completed] 11 with both the electronic Cash Account 12 information retrieved and also copies of the 13 paper cash accounts for week 42 (and also weeks 14 41 and 43) ... Specifically, the Carried Forward 15 figures from Week 41 matches the Brought Forward 16 figures in Week 42 and the Carried Forward 17 figures from Week 42 matches the Brought Forward 18 figures in Week 43. 19 "The initial check was that each Session's 20 data was completely recorded in the Transaction 21 Log. This check identified a transaction 22 missing from the ARQ 421 data for a value of 92p 23 on 12 January." 24 That's referring to the issue you picked up 25 in the paragraphs we have just looked at? 26 1 A. Yes. 2 Q. "This transaction did not include its Start Time 3 (a known fault that occasionally happens) and so 4 the ARQ extraction process ignored it. However 5 it would not have been ignored by the accounting 6 functions at the counter and a report would have 7 been generated that night as part of the 8 overnight checks. 9 "Unfortunately, this report is not audited 10 and so is not available for examination. 11 However, we do not believer that this report is 12 material to the case." 13 So the idea of doing the three checks that 14 you mentioned seems to have been dropped, 15 doesn't it? 16 A. I don't think so. I cannot remember but it 17 could well be that, by the time that Gareth 18 wrote this final version, he had checked with 19 the audit team and they had said, "No, that file 20 is not audited or available anywhere", which is 21 what -- that's what he's saying there. 22 I wouldn't have -- I don't know. 23 The other checks -- 24 Q. Just hold on. On what basis were you saying 25 that -- you are now saying that this report, 27 1 which you'd previously wished to see, is not 2 material to the case? 3 A. I don't know quite why Gareth put that sentence 4 in there. 5 Q. Did you see this report before it went out? 6 A. I probably did. I mean, it -- I don't think it 7 would have shown any further light on the -- any 8 of the losses or anything because all the report 9 would have shown was that a single transaction 10 with a missing start date had been picked up on 11 that day, and we already knew from the full set 12 of data that, yes, there was a transaction with 13 the missing start date on that day. 14 Q. You'd previously made three suggestions. 15 A. Yes. 16 Q. It appears that two of them hadn't been pursued. 17 A. I don't think that's -- I mean, I don't think 18 there's any -- anything in there that suggests 19 that the other two were not pursued. 20 Q. Why were you -- why was it the case that they 21 were -- the results of those checks, if they had 22 been conducted, are not reflected in this 23 document? 24 A. Because -- I think because we found nothing, so 25 there was nothing to say. I obviously cannot 28 1 now prove that I did those other things but, 2 knowing how thoroughly I went into everything, 3 I think it's very likely that I -- sorry, I now 4 can only remember what one of the checks was but 5 I was checking everything that occurred to me. 6 If I'd checked but found nothing else of 7 concern, then that might not have been recorded 8 in this document. 9 Q. Thank you. That document can come down. 10 We've heard evidence from your then line 11 manager Mik Peach that before you were asked or 12 requested to provide evidence in the Lee 13 Castleton case, another person at Fujitsu within 14 the Security team, whose function or 15 responsibility it was ordinarily to give 16 evidence in a case such as Mr Castleton's, had 17 declined to give evidence at court. 18 A. I -- 19 Q. I'm just summarising what Mr Peach has told the 20 Chairman? 21 A. I had no knowledge of that. 22 Q. I was about to ask. Did you know that another 23 person within the Security team had declined to 24 give evidence -- 25 A. No. 29 1 Q. -- in Mr Castleton's case -- 2 A. (The witness shook her head) 3 Q. -- and, therefore, you couldn't help us as to 4 who that person was or why they had declined to 5 give evidence? 6 A. No, I couldn't. 7 Q. Mr Peach told the Chairman that you were 8 pressured to give evidence and to go to court, 9 and that you were pressured to give evidence and 10 go to court because the person from the Security 11 team had refused to give evidence and to go to 12 court. Were you pressured into giving evidence? 13 A. I'm not sure I was pressured; I was extremely 14 surprised to find myself in a position where 15 I had to, because it had never occurred to me or 16 any of my colleagues that this might be part of 17 our role supporting an IT system. 18 I just felt that it was all sort of handled 19 rather strangely and that I was asked by this 20 guy I didn't know, who turned out to be -- who 21 was the Security Manager, if I'd be prepared to 22 speak to a solicitor, and I -- about this 23 particular matter and I said, "Well, I suppose 24 so", and there was no mention at that point 25 that, just by doing that, I would end up giving 30 1 evidence in court and here today. 2 Q. Somebody from Security asking you whether you 3 would mind speaking to a solicitor -- 4 A. Mm. 5 Q. -- and you saying, "Yes, I suppose so" -- 6 A. Mm. 7 Q. -- doesn't sound like pressure? 8 A. It just seemed to be an odd situation because 9 then my manager was very cross about it all and 10 I didn't know if it was me having said yes to 11 that initial question, somehow had led to 12 everything else that happened and, if I'd said 13 no, would it have been different? I don't think 14 that is the case now but now I know a lot more 15 about the background of it. 16 But, at the time, I just felt very 17 uncomfortable about it and there seemed to be 18 a lot of people not very happy with the 19 situation and so, yes, it was just 20 uncomfortable. I don't think I was ever given 21 the opportunity to say "No, I'm not prepared to 22 do this", and I don't know what would have 23 happened if I had turned around and said no. 24 I've always, you know, attempted to do my job 25 and to be helpful and to answer questions as 31 1 necessary. 2 Q. Were you aware at the time that Mr Peach, your 3 manager, had an argument about you having to 4 give evidence, in a corridor with either Dave 5 Baldwin, the Customer Services Director and 6 Naomi Elliott, the Support Services Manager, or 7 Brian Pinder from the Security team? 8 A. I wasn't aware of it at the time but it doesn't 9 surprise me. 10 Q. Why doesn't it surprise you? 11 A. Because Mik was very angry about one of his team 12 being put in this position that he thought they 13 should not have been put in. 14 Q. So the argument wasn't reported back to you at 15 the time? 16 A. I -- not that I'm aware of, no. I know he was 17 talking to his management about this. 18 Q. Mr Peach told us that, following the argument, 19 he had a choice to select anyone from within the 20 SSC to give evidence and that he chose you 21 because you were the most experienced and 22 technically best within the SSC in the relevant 23 area of counter code that he -- you had dealt 24 with a call back on 26 February 2004 and he had 25 confidence in your honesty and integrity. That 32 1 makes it sound like he is picking you? 2 A. Yes, I wasn't aware of that. I assumed it was 3 me because I'd looked at the call in 2004. 4 Q. Can we look, please, at your witness statement, 5 please, at page 18, and paragraph 56. If we 6 scroll down, please -- and a bit more, thank 7 you. 8 If we just read these, from 56 to 59 to see 9 the account that you give: 10 "I have not been involved in any other 11 proceedings against subpostmasters involving the 12 Horizon IT System before or since the Castleton 13 case. Before 2006 I had never taken part in any 14 legal proceedings of any kind. It had never 15 been suggested to me or my SSC colleagues that 16 we might be expected to become involved in such 17 proceedings as a result of the work that we were 18 doing. 19 "On a few occasions over the subsequent 20 years Gareth Jenkins asked me to double check 21 his figures or help him locate KELs or reports. 22 I was aware he was giving evidence in various 23 trials but I do not think I ever knew the 24 details of those trials and the assistance 25 I provided to him would have been limited. 33 1 I have no specific recollection of any 2 particular instances of this occurring. 3 "I am asked who asked me to provide 4 a witness statement in the case against 5 Mr Castleton. Sometime in 2006, a man I did not 6 know came into our secure office area, 7 introduced himself as the Post Office Account 8 Security Manager, Brian Pinder, mentioned a call 9 I had dealt with over two years earlier and 10 asked if I would mind having a quick chat with 11 a solicitor about it. No mention was made of 12 a witness statement at that point nor, as 13 I recall, of a court case. 14 "I informed my manager [Mr Peach] of this 15 conversation. He was extremely unhappy that 16 I had been approached directly and said that no 17 member of SSC should be involved in litigation. 18 He then talked to higher levels of management, 19 but my name was now in the frame and I seemed to 20 have no option but to talk to the solicitor 21 which led to them requesting a witness statement 22 and my subsequent appearance in court." 23 You see that you got it another way round, 24 namely Security, through Mr Pinder, approaching 25 you, then you complaining to Mr Peach and then 34 1 him raising it with higher levels of management, 2 rather than management speaking with him and him 3 choosing you. 4 A. That was my -- how I thought it was, until 5 I read or saw what Mik said in evidence to the 6 Inquiry. 7 Q. Has your recollection altered at all in the 8 light of what Mr Peach has said to the Inquiry? 9 A. No, I don't think so, except possibly his 10 reaction now is a little bit more 11 understandable. 12 Q. Your recollection is they came direct to you, 13 then you went to Mr Peach? 14 A. Yes, Brian Pinder most definitely came and 15 talked to me and then I mentioned it. So 16 whether -- I'm assuming now that Mik knew that 17 they were likely to want to talk to me but he 18 was probably very cross then that they had 19 actually come straight to me, rather than go 20 through him. 21 Q. Can we just look at your afterthoughts document 22 because this I think helps us a little bit on 23 this issue. FUJ00152299. 24 We're going to look at this in more detail 25 a little later this morning but if we just turn 35 1 to page 2 to start with, we can see that this is 2 a document that you wrote on 29 January 2007. 3 A. Yes. 4 Q. So it was prepared much nearer to the events 5 we're considering? 6 A. Yes. 7 Q. We're going to look, as I say, in detail at this 8 document a little later. But I just want you, 9 if we go back to page 1, please, to look at 10 what's said in the first paragraph: 11 "In the summer of 2006 I was asked directly 12 by the Security Manager ..." 13 That's Mr Pinder, yes? 14 A. Yes. 15 Q. "... whether I would be prepared to speak to 16 a solicitor about a call I had dealt with in 17 February 2004. My initial response was that 18 this was not the normal process, but he 19 reassured me that it was more or less 20 a formality so somewhat reluctantly I agreed." 21 That is consistent with what you told us and 22 you said in your witness statement that it was 23 a direct approach to him -- 24 A. Yes. 25 Q. -- rather than through Mr Peach? 36 1 A. Yes. 2 Q. You say that you said this was not the normal 3 process. What was the normal process? 4 A. I'm not sure I knew precisely what the normal 5 process was. I'm not sure if there had been any 6 previous court cases. I think I was aware that 7 some people within the Security team, perhaps in 8 particular Penny Thomas, would occasionally have 9 to go to court to give evidence. 10 Q. Just interrupting you there, if I may. Did they 11 liaise with the SSC in preparation to give 12 evidence to court? 13 A. I've got no specific memory of that happening. 14 I don't think I knew Penny very well at that 15 stage and I don't recall having actually worked 16 with her on anything. 17 Q. What about Andrew Dunks, can you remember him 18 performing the same function? 19 A. Yes, again, I mean, I knew who he was but 20 I don't think I'd had anything specifically to 21 do with him. Possibly -- and I can't remember 22 if this is before or after -- he might have 23 asked me about a particular PowerHelp call or 24 something but I don't think I'd done -- I'm sure 25 I had not done anything, except possibly answer 37 1 a sort of very quick, simple question. 2 Q. Were you aware of any of those people, 3 ie Mr Dunks or Penny Thomas or similar from the 4 Security team, obtaining evidence or material 5 from the SSC, as part of the process of 6 preparing to go to court? 7 A. I don't remember that happening. I think the 8 Security team were very separate and they had 9 access to the ARQ data and the -- well, it was 10 their job to extract the ARQ data and -- 11 Q. Were they experts in its interpretation? 12 A. No. 13 Q. Okay, that can come down, please. If we can go 14 back to paragraph 61 of your witness statement, 15 which is on page 19. Page 19, please. 16 At the foot of the page, paragraph 61, you 17 say: 18 "I think there was an initial meeting with 19 a solicitor, presumably Stephen Dilley, in 20 a Fujitsu office in Bracknell. Mik Peach's 21 manager, Naomi Elliott, was present at that 22 meeting or a subsequent one. It was explained 23 to me, I think in the initial meeting, that 24 I would be a Witness of Fact not an Expert 25 Witness, and the purpose of my witness statement 38 1 was to document my investigation of 26 February 2 2004." 3 A. Yes. 4 Q. You recall that you were told in the first 5 meeting that you would be a Witness of Fact -- 6 capital W, capital F -- not an Expert Witness -- 7 capital E, capital W. Did you then have 8 an understanding of the distinction? 9 A. Not clearly. I've got very little memory of 10 this meeting but I know -- I'd heard the phrase 11 "Expert Witness" before but I don't think I'd 12 ever heard the phrase "Witness of Fact" but it 13 was explained to me that, you know, I wasn't 14 going there to talk about the whole system. It 15 was just to say about my investigation and what 16 I had found. 17 Q. Who gave you this explanation? 18 A. I can't remember. 19 Q. Was it -- 20 A. Most likely, I think the solicitor but I suppose 21 it could have been Brian Pinder or somebody, but 22 it seems more likely that it was Stephen Dilley. 23 Q. By this time, had you had any training or 24 guidance from Fujitsu or, indeed, the Post 25 Office on the differences between those two 39 1 broad species of witness? 2 A. No training or guidance. Just this sort of 3 basic statement that I was just talking about 4 what I had done and not about the overall 5 system. 6 Q. Had you received any training or guidance from 7 Fujitsu or, indeed, the Post Office on the 8 responsibilities involved in giving evidence as 9 a Witness of Fact or as an Expert Witness? 10 A. I don't recall anybody spelling out any specific 11 responsibilities. 12 Q. Now, subsequently, we know that you carried out 13 investigations and gave some evidence about 14 matters beyond the strict confines of what you 15 had done on 26 February 2004? 16 A. Yes, I did. 17 Q. You had analysed with Mr Jenkins the contents of 18 the solicitor's letter and provided a response 19 to it? 20 A. Yes, I think at the time I saw that as being 21 a sort of separate exercise, not connected 22 really with the witness statement that I was 23 making. 24 Q. But you got more dragged into things, other than 25 the strict confines of the facts of what had 40 1 happened on 26 February? 2 A. Yes, I mean, partly through my own choice 3 because I wanted to be absolutely sure as 4 I could be that I hadn't missed anything at all 5 back in that original February. As we discussed 6 yesterday, I hadn't possibly gone back as far as 7 I could have done in January. So, in the run-up 8 to the trial or so -- and so on, I spent a great 9 deal of time rechecking the archived audited 10 message store data to see if I could find any 11 reason for the losses that I might have missed 12 originally. 13 Q. You, in the witness statement and then in court, 14 gave a view on that; you gave your assessment? 15 A. I'm not sure that my witness statement said 16 anything about any subsequent checks that I had 17 made and I'm not sure that that actually got 18 picked up on in court either. 19 Q. Had anyone told you about the need -- if you are 20 conducting investigations, where you may end up 21 giving a view or an assessment -- of the need to 22 make a record, contemporaneous record of what it 23 is you're doing -- 24 A. No. 25 Q. -- and to retain the product of your 41 1 investigations -- 2 A. No, nobody ever said anything like that? 3 Q. -- your working notes or the data that you'd 4 captured? 5 A. No. I mean, for the -- looking at the CAP42 6 cash account, I know we do have that data 7 because we have it here now -- and, obviously, 8 I kept that and then passed it on to Gareth so 9 he could check, and so on. But all the other 10 checks I made, you know, I would have had a lot 11 of various spreadsheets, and so on, at the time 12 but nobody said I needed to retain them or say 13 that I'd got them. 14 Q. You say, subsequently, that you were put in 15 an invidious position when you revealed in court 16 that you had used the Tivoli event log in the 17 course of your investigations and that hadn't 18 been disclosed? 19 A. That I'd used them in the 2006 investigations, 20 yes. Yes. 21 Q. You regarded yourself as -- we're going to look 22 at it in the afterthoughts document in 23 a moment -- being in a difficult spot because 24 you were talking about something and there had 25 been no disclosure of the underlying material? 42 1 A. Yes. I mean, yes, we might want to talk about 2 this some more but, yes, I felt very awkward 3 about that. 4 Q. Was there a discussion at the time about the 5 awkwardness that you felt about being on the 6 stand in the High Court, speaking about events 7 in respect of which there had been no 8 disclosure? 9 A. I'm -- discussion with whom? 10 Q. The solicitor, Mr Dilley? 11 A. I -- I'm not sure that it was discussed. Oh, 12 I know Stephen Dilley did subsequently send me 13 an email, which I then passed on to various 14 other people in Fujitsu, pointing out the 15 disclosure rules, and so on. But I personally 16 had no responsibility for disclosures, and so 17 on, and hadn't -- you know, didn't know what had 18 been disclosed or what should be. 19 These particular event logs, we'd only found 20 that they did still exist about a couple of 21 weeks before the trial started, when -- 22 following up a different query that somebody had 23 made regarding Marine Drive, and it was just 24 because one of my colleagues in SSC went looking 25 to see what else -- whether these files were 43 1 archived, and discovered that, yes, they were. 2 It wasn't part of the formal audit process; 3 it was just a file archive. But yes, they 4 existed and so, when I found that I was talking 5 about them because of a misunderstanding -- 6 Q. Between you and Mr Morgan? 7 A. -- between me and Mr Morgan and I was asked if 8 they existed, of course I said "Yes, they do". 9 Q. And that caused a kerfuffle? 10 A. It felt like a little bit of a kerfuffle because 11 I think everybody was ready to wind everything 12 up and go away for Christmas -- 13 Q. And you had to come back in January? 14 A. -- they had to come back again. 15 Q. Is this a fair summary: although you spent time 16 in 2006 looking at a wider range of data than 17 you had looked at in 2004, you didn't keep 18 a disclosable record of what you did in 2006, 19 what record you used, what you were looking for 20 and what the product of it was -- 21 A. Yes. 22 Q. -- is that fair? 23 A. I didn't keep that at the time, no. 24 Q. So Mr Castleton and his legal team, if he had 25 one, had no way of reviewing what you had done, 44 1 replicating it and finding out for themselves 2 whether there were mistakes in your approach or 3 your methodology? 4 A. No, although I was looking really to see if 5 I could find something that would help him, 6 rather than the opposite. 7 Q. Yes, I completely understand that. 8 A. Yes. 9 Q. But, in the event, the material to show your 10 workings out was not available? 11 A. That was not available, no. 12 Q. Similarly, when in court you spoke about the 13 Callendar Square bug -- 14 A. Yes. 15 Q. -- the relevant underlying documents, for 16 example the KELs, hadn't been provided. It was 17 just your evidence about what had happened with 18 the Callendar Square bug and your view that it 19 hadn't afflicted Marine Drive that the court had 20 to go on? 21 A. Yes. I had been told quite early on in the 22 process that KELs was something that were not 23 disclosed. 24 Q. Why were KELs not disclosed -- 25 A. I don't know. 45 1 Q. -- and who told you this? 2 A. I can't remember if it was my manager or 3 somebody on the Security team but I was told 4 explicitly to talk about PowerHelp calls, rather 5 than PEAKs and not -- and that KELs weren't 6 disclosed nor PEAKs. 7 Q. Was that because they revealed known errors, 8 from their title? 9 A. I don't know why. I was just told that was how 10 it was. 11 Q. You said one of your managers; does that mean 12 Mr Parker or Mr Peach? 13 A. At that point, it would have been -- if it was 14 my manager, it would have been Mik Peach or, 15 potentially, Naomi Elliott. 16 Q. Ie the manager of the manager? 17 A. Yeah, or potentially Brian Pinder or potentially 18 Stephen Dilley but -- no, I don't think that was 19 Stephen Dilley, I think it was somebody on the 20 Fujitsu side. 21 Q. So Known Error Logs just weren't disclosed and 22 that was the way it was? 23 A. That is what I was told and, although I thought 24 it seemed strange, I didn't know it was wrong. 25 Q. Why did you think it was strange? 46 1 A. Because I would have thought -- you know, in my 2 layman's terms, I thought you had to disclose 3 everything. But I was not a legal expert and 4 I had no idea, you know, whether -- as I said, 5 whether this was right or wrong and it was 6 not -- you know, not my responsibility. 7 Q. This was in the context of a case where a man 8 was saying, "I think there are things wrong with 9 the system, I, Mr Castleton", and these known 10 error logs revealed problems with the system, 11 whether they were attributable to his branch or 12 not. That was in the context you were thinking, 13 presumably? 14 A. Um, no, it was just sort of a general -- no, 15 "There is this information, do we not have to 16 share it with them?" 17 Q. You said that you were told to refer instead to 18 the PowerHelp records? 19 A. Yes. 20 Q. The PowerHelp records are not a replica of the 21 Known Error Logs, are they? 22 A. No. 23 Q. There's no equivalence between them? 24 A. No. 25 Q. In particular, the Known Error Log might reveal 47 1 for how long a problem was known by Fujitsu and 2 the promptitude of steps taken to resolved it? 3 A. Yes, you couldn't necessarily work out the 4 timescale from the KEL, for various reasons, but 5 it would give you more of a clue as to what was 6 happening and certainly when it started. 7 Q. You'd be more likely to be able to work out when 8 the problem first emerged? 9 A. It would probably give you more of a clue, yes. 10 Q. There is more discussion in a KEL, is this 11 right, more inward facing discussion by Fujitsu 12 as to whether the error is acknowledged, than in 13 a PowerHelp call, which is more about a customer 14 complaining? 15 A. Yes. I think when there were errors we 16 acknowledged them, certainly on PEAKs, but 17 PowerHelp calls, it just depends who dealt with 18 them and who decided to put information on them. 19 Q. It could be very variable the information that 20 went onto a PowerHelp -- 21 A. Yes. 22 Q. -- but certainly the customer complaint, the 23 branch saying, "I've got a problem"? 24 A. Yeah, yeah. 25 Q. But it was hit and miss whether the 48 1 investigation and the acknowledgement of 2 a problem got onto a PowerHelp call? 3 A. To some extent. I mean, I think these calls 4 that you have been looking at are the ones where 5 that was most likely to happen. I would say the 6 vast majority of PowerHelp calls, which were 7 much more fixed issues than -- yeah, they were 8 pretty good. But yeah, these areas where the 9 responsibility between NBSC and -- 10 I think -- yes, if there was a known problem 11 identified by PowerHelp, then, certainly if it 12 had been to -- had come to PEAK, you would have 13 had the PEAK response on there too. If it had 14 just been handled by the Helpdesk, then they 15 might not have spelt it out so much, if it was 16 a system error. 17 Q. So you ended up giving evidence about the 18 Callendar Square bug -- 19 A. Yes. 20 Q. -- without the two KELs having been disclosed 21 that revealed that there had been an appreciable 22 delay in addressing the Callendar Square bug? 23 A. Yes, those KELs weren't disclosed. 24 Q. Looking back, do you feel that the initial brief 25 that you were going to give evidence, factual 49 1 evidence, about your limited Helpdesk 2 investigation conducted within a five-hour 3 period on 26 February 2004 was honoured? 4 A. No. Partly because Mr Castleton asked me 5 much -- a much wider range of questions than 6 I was expecting but it seemed only right to try 7 to answer the questions that were put to me. 8 But at times I realised I sounded a little 9 unsure because it wasn't things that I had been 10 expecting to be asked about. 11 MR BEER: Thank you. 12 Sir, that's an appropriate moment for the 13 morning break, if it's convenient to you. 14 SIR WYN WILLIAMS: Yes, certainly. So what time 15 shall we resume? 16 MR BEER: 11.30, please. 17 SIR WYN WILLIAMS: Fine. 18 MR BEER: Thank you. 19 (11.15 am) 20 (A short break) 21 (11.30am) 22 MR BEER: Good morning, sir. Can you see and hear 23 me? 24 SIR WYN WILLIAMS: Yes, I can, yes. 25 MR BEER: Thank you very much. 50 1 Mrs Chambers, can we see what you say in 2 your witness statement about a couple of issues 3 that concern the topic we're presently on and, 4 therefore, if we can look at page 23 of your 5 witness statement. At the foot of the page, 6 paragraph 74, you say: 7 "On 17th August 2006 I wrote to Gareth in 8 the following terms: 9 "I've spent some time recalculating the 10 CAP42 cash account for Marine Drive and 11 addressing the points in the letter from the 12 solicitor. I hope it makes enough sense to 13 provide you with a starting point not quite at 14 the very beginning. Subsequent cash accounts, 15 if needed, should be much easier now I have the 16 mapping is set up and some idea of what numbers 17 need to go where." 18 Then scrolling down: 19 "I'm now going to look at the loss made in 20 week 42 and demonstrate that it was due to the 21 difference between system holdings and declared 22 holdings. 23 "If this isn't at all what you wanted, 24 please let me know -- I don't really know what 25 I'm doing! 51 1 "This hasn't had my full attention, lots of 2 people are on leave and Martin landed me with 3 a tricky POLFS/FP issue. Also, yesterday I got 4 my witness statement which is (as I expect you 5 found) full of things I didn't say or do, 6 including all those PowerHelp calls." 7 Picking up on a couple of things that you 8 said there, you say that you are looking to 9 demonstrate that the loss was due to the 10 difference between system holdings and declared 11 holdings. 12 A. Yes. 13 Q. Would that exercise tell us if the holdings 14 recorded by the system were inaccurate by reason 15 of a bug, error or defect in Horizon? 16 A. Sorry, I need to think about that a little bit. 17 (Pause) 18 The system holdings, you calculate -- well, 19 you can see the opening figures at the start of 20 a period and then by looking at all the 21 transaction data that has affected the different 22 products, and so on, you can check whether 23 the -- what the system holding should be at the 24 end of the next period, for example. And, 25 obviously, if that calculation shows a mismatch 52 1 between, you know, what you calculated it should 2 be and what the system has calculated it should 3 be, then that would be a system problem. 4 But if they are the same, then, you know, 5 it's a matter of arithmetic, not opinion, and if 6 the system holding has been calculated correctly 7 and it's not the same as the declared holdings, 8 then, yes, you have a loss or a gain, which 9 would appear not to be the fault of the system. 10 Q. You were adding up what you could see in the two 11 records, rather than undertaking any probing 12 investigation into whether there may be 13 a problem with the recording of the transactions 14 on Horizon? 15 A. I was adding up the transaction data that had 16 been recorded, which was all I had to go on. 17 I mean, there were, obviously -- I'd checked for 18 things like everything netting to zero, which 19 had been okay, apart from the two things that 20 had been -- the one line identified. 21 I knew from the ARQ extract that part of 22 that checking made sure that the set of 23 messages -- there were no gaps in the messages 24 received from the branch. So everything that 25 had been written to the message store on the 53 1 counters appeared to have been included in the 2 data that I was now doing this comparison with. 3 Q. This was essentially an accountancy exercise? 4 A. It was essentially an accountancy exercise, yes. 5 Q. Adding up totals -- 6 A. Yeah. 7 Q. -- in different sets of data with the 8 assistance, presumably, of some Excel? 9 A. Yes. 10 Q. You say, if we scroll down, please, at 11 paragraph 76: 12 "I thought at the time, and still think, 13 that this was a job for an accountant." 14 Yes? 15 A. Yes. 16 Q. It didn't involve any computer expertise -- 17 A. Not really, no. No. Obviously not -- 18 Q. -- or expertise in computers? 19 A. Yes, that's true. 20 Q. So your investigation didn't extend to whether 21 there was a problem with the recording of the 22 transactions beyond the extent that you've said? 23 A. There was no indication of any problem with the 24 recording of the transactions that was visible 25 to me, either when I looked in 2004, when 54 1 obviously there was, you know, more files and 2 things to look at and, okay, it was only for 3 a shorter period, and I couldn't see anything 4 wrong with the recording of the transactions 5 subsequently. 6 It would only be by checking somehow, 7 checking against what the branch had actually 8 intended to record that you could see if there 9 was anything wrong in what had been recorded. 10 There were no gaps in the sequence but 11 whether -- certainly, it wasn't that individual 12 transactions had been dropped, if that had 13 happened, then you would have receipts and 14 payments mismatches. We didn't have those. 15 Whether entire sessions had not been 16 recorded, that I could not see. And whether 17 some of the sessions that had been recorded were 18 not as intended at the branch, I also could not 19 see that, without some way of knowing actually 20 what had happened at the branch. 21 Q. One way of doing that would be to send somebody 22 in on balancing day, for example -- 23 A. Yes, or just during normal processes. 24 Q. -- and just watch the subpostmaster or their 25 clerk do it? 55 1 A. Yeah, and try to keep a record that you could 2 check against at the end of the day. I mean, 3 the postmaster had a lot of reports that had to 4 be printed out at the end of the day, with 5 totals on for pensions and various other things, 6 and I believe that -- but this is getting into 7 business stuff, which wasn't -- I had less 8 familiarity with, but they were meant to add up 9 the dockets or counterfoils, or whatever they'd 10 got, for various things and compare them against 11 the totals on the reports, to make sure that 12 what was on the system was consistent with the 13 business that they had done. 14 But that was something that I had no way of 15 cross checking. 16 Q. Those are two things that could be done to seek 17 to discover whether there was an underlying 18 problem and, if so, what it was? 19 A. Absolutely, and it is possible that if those 20 sort of checks had been done, it might have 21 highlighted some sort of system problem. At the 22 time, my view was that seemed very unlikely, 23 but -- or, you know, completely unlikely, 24 completely impossible, but, in the light of 25 where we are now, who knows. 56 1 Q. Because there's some system faults that are not 2 visible to you, sitting in an office in 3 Bracknell? 4 A. No, not for that reason. It would be a fault 5 that we still haven't discovered. 6 Q. I see. An unknown system fault -- 7 A. An unknown, unknown -- 8 Q. -- that doesn't obviously leave a trace? 9 A. Yeah, that doesn't leave any trace and that 10 nobody notices happening at the time, except 11 somehow you have a loss at the end of the week. 12 Q. The two things that we've just discussed as next 13 steps, are those the kind of things that you 14 imagined the Post Office might do? 15 A. Yes, I hoped that Post Office had staff who were 16 very familiar both with the business processes 17 and had a good working knowledge of the Horizon 18 System as well. 19 Q. The passage of -- if we just scroll up, please, 20 to the middle of page 24, please, where you say: 21 "If this isn't at all what you wanted, 22 please let me know -- I don't really know what 23 I'm doing!" 24 What did you mean by "I don't really know 25 what I'm doing!"? 57 1 A. This was the first time I'd tried to calculate 2 a cash account by hand. We've seen from looking 3 at them they're fairly horrible, pages and pages 4 of stuff, and I was much more familiar at 5 looking at the trial balance and final balance 6 reports, which lay out the accounts, to my mind, 7 in a much more intuitive way. And then the same 8 data got reanalysed using different mappings, 9 that's saying which lines go on different cash 10 account, to produce this document, that then, 11 I believe, had to be signed off and went to Post 12 Office. 13 And I was much less familiar with how the 14 cash account processing actually allocated 15 things to particular lines. 16 So I was really having to work that out from 17 scratch, and I -- although I was -- thought 18 I was probably on the right lines, Gareth was 19 much more of an expert in this area, so I really 20 didn't want him just taking what I'd done 21 without checking it pretty thoroughly. 22 Q. Thank you. Can we move ahead, please, to 23 POL00069622. You attended a conference on 24 11 September 2006, a conference meaning 25 a meeting with lawyers and witnesses -- 58 1 A. Yes. 2 Q. -- and this is an attendance note in respect of 3 it? 4 A. Yes. 5 Q. Mr Dilley records: 6 "Had a conference with Counsel, Richard 7 Morgan and Tom Beezer, Partner Bond Pearce at 8 Counsel's Chambers in London." 9 Do you remember going up to London to 10 counsel's chambers. 11 A. I remember going up to London to counsel's 12 chambers, yes. 13 Q. We can see the purpose of the conference is 14 recorded to meet four of the key witnesses to go 15 over their draft statements with them. 16 A. Yes. 17 Q. The conference note records them one by one, 18 starting with Cath Oglesby, yes? 19 A. I don't believe we were all together at the same 20 point. I think we were there separately. 21 Q. If we scroll forwards, please, to page 4 at the 22 bottom, and scroll down, please. We see at the 23 foot of this page the record concerns you: 24 "Meeting with Anne Chambers. 25 "We went through Anne's Witness Statement. 59 1 She said she had personally got a new set of 2 referrals for six years." 3 What does that mean? I don't really 4 understand it. 5 A. I have no idea. 6 Q. Then this: 7 "Three to four years subpostmasters had been 8 complaining that there is a problem or have 9 complained if there is a problem. Sometimes 10 there is a major [blank] for example, all the 11 cash and stock appears to have vanished out of 12 the office. But these sort of errors are 13 singular and not continual." 14 Would it be right that at this time, so 15 autumn 2006, subpostmasters have been 16 complaining for three or four years that there 17 were problems with the Horizon System? 18 A. Yes, I think that was probably true. I wouldn't 19 say that there were large numbers of these sort 20 of complaints but certainly there were some 21 concerns, I think. 22 Q. So remembering as best you can, you would have 23 said, because it would have been accurate, that 24 for three or four years subpostmasters had been 25 complaining about Horizon, rather than three or 60 1 four postmasters had reported complaints? 2 A. Yes, I think that does mean that, over the -- 3 yes, three or four years, rather than three or 4 four postmasters. 5 Q. It would have been wrong to say that only three 6 or four postmasters had made complaints about 7 Horizon? 8 A. Yes, I think so. You know, it wasn't -- 9 certainly wasn't up in the hundreds but it was 10 probably more than three or four. 11 Q. You say or it is recorded that you said: 12 "Sometimes there is a major [then the word 13 is missing] for example ..." 14 Can you help us as to what that might be, 15 looking at the context, the missing passage? 16 A. A major problem, I don't know. I think one of 17 the documents I saw very recently appears to be 18 another bug, error or defect that I had totally 19 forgotten about. When -- and it was probably 20 starting up -- started happening around about 21 the time that I was having this meeting, which 22 was probably why it was in my mind, where -- 23 I can't remember the details but, yes, they 24 rolled over and lost their stock. 25 But that was -- it was so obvious that 61 1 something had gone horribly wrong, if you like, 2 that I believe that got picked up on and 3 investigated. It wasn't something that was 4 affecting many branches and I think it was those 5 that were doing "declare stock", which was not 6 what Mr Castleton did anyway. 7 Q. Was this pursued with you in any way, what you 8 said here, that for three to four years 9 subpostmasters had been complaining about 10 Horizon and that sometimes there are major 11 problems, namely cash and stock appears to have 12 vanished out of the office? 13 Questions about what are the problems, how 14 many of them are there? How many subpostmasters 15 have been affected? How are they recorded in 16 the SSC? Can we get access to documents? Those 17 kind of questions. 18 A. I don't recall any of those kind of questions. 19 I think I was just being asked, generally, "Are 20 there problems with Horizon?" 21 Q. And you said yes? 22 A. And I gave a general answer and then I gave 23 a specific example of something that was 24 happening recently and also pointed out that, 25 you know, these weren't things that were 62 1 affecting large numbers of branches, to my 2 knowledge. 3 Q. So would it be fair to say that, in this 4 conference, the nature and extent of the Horizon 5 problems was not explored in any detail? 6 A. I think that's true. I mean, to my view, at the 7 time, we were talking about Marine Drive, which 8 I was very confident had not had any of these 9 particular problems. 10 Q. Can we look, please, at your reflections 11 document. That can come down and instead look 12 at FUJ00152299. 13 We looked at this earlier, do you remember, 14 29 January 2007? 15 A. Yes. 16 Q. There are four headings. 17 A. Yes. 18 Q. "Approach to SSC staff", "Review of technical 19 evidence", "Disclosure of evidence" and then, 20 over the page, "Helpdesk calls". 21 A. Yeah. 22 Q. Did that reflect four issues that you had 23 identified as part of this entire process of 24 being asked to give witness evidence and then 25 give evidence in court? 63 1 A. Yes. These were issues that I obviously felt 2 concerned enough about to feel that, you know, 3 having come to the end of this process, 4 I thought, that I ought to -- I was expecting 5 there to be some sort of investigation -- some 6 sort of wrap-up, and I felt it was important to 7 get these things down. I had completely 8 forgotten, until this was disclosed to me again, 9 that I had written this document. But yes, 10 I obviously -- very obviously did and I am quite 11 glad I did. 12 Q. Yes. If we go back to the first page, please, 13 the four topics, did they reflect quite 14 significant concerns that you had, having come 15 to the end of the process? 16 A. Yes. 17 Q. We've already looked at the first paragraph, 18 "Approach to SSC staff". Can we look at the 19 second and third paragraphs under that -- no, 20 sorry, the second and third paragraphs under 21 topic 1. Thank you. You say: 22 "Subsequently, before the meeting with the 23 solicitor, he asked me what my availability was 24 in the autumn for the court case. This was the 25 first time there was any mention of the 64 1 possibility of having to go to court. Repeated 2 assurances that this would all be settled before 3 getting to court proved to be unfounded. 4 "I appreciate there may be circumstances 5 where witnesses are summoned and have no option 6 but to comply, but I was not at all happy about 7 how this was handled." 8 On that issue, were you expecting something 9 to be done as a result of your raising this 10 issue to regulate the circumstances in which SSC 11 staff were approached to provide evidence for 12 use in court proceedings? 13 A. Yes, I think I felt that if this was part of SSC 14 members' role, then that should be made very 15 clear to anybody wanting to join SSC. 16 Q. Anything beyond that, rather than just telling 17 people "By the way, when you join -- when you're 18 one of the 25 or 30 of us, you might end up in 19 court giving evidence". Did you have anything 20 further in mind about regulating how people were 21 approached, in what circumstances they were 22 approached and bringing some formality to bear? 23 A. Yes, I think all of that and possibly rather 24 more training and guidance than I felt I had. 25 Q. So that was your hope? 65 1 A. Yes. 2 Q. Can we look at your second concern, "Review of 3 technical evidence". You say: 4 "When I took the initial call in February 5 2004, I only spent a few hours on it before 6 deciding that could not see any sign of a system 7 problem. I only looked at a couple of week's 8 information." 9 That probably helps us with some of the 10 answers you were giving yesterday. 11 A. It does, yes. 12 Q. So a few hours would have been within that 13 five-hour window and to you only looked at 14 a couple, presumably meaning one or two? 15 A. Yes, again, I'm writing this two and a half 16 years later, so my memory of exactly what I had 17 done was no better than it was in my witness 18 statement. 19 Q. You say: 20 "While in this case I am now sure that I did 21 not miss anything, and my initial analysis was 22 correct, I am concerned that there was no 23 technical review of the Horizon evidence between 24 the original call and the case going to court. 25 It is probable that any system problem affecting 66 1 the accounts would have to be shown up to Post 2 Office staff who did check the figures very 3 carefully, but since the postmaster was blaming 4 the system for the losses I think it would have 5 been sensible to have double checked this with 6 Fujitsu before it got as far as court. I was 7 certainly concerned, in the early stages, that 8 there might be something I had missed." 9 Just stopping there. Were you essentially 10 suggesting by that that, before court 11 proceedings are launched, the Post Office should 12 come back to Fujitsu for a check of some kind on 13 the data? 14 A. It would seem very sensible, if you want to get 15 to the bottom of somebody's problems, if the 16 consequence of those problems means that they 17 may be sent to prison. 18 Q. So your hope was that something would be done as 19 a result of you raising this suggestion? 20 A. I felt it really needed to be said. 21 Q. You knew: 22 "Once in court, I found myself being treated 23 as an expert witness and answering a wide 24 variety of questions about the system, although 25 nominally I was a witness of fact and my witness 67 1 statement just covered the investigation done in 2 2004. Fortunately I do have extensive knowledge 3 of the system and was able to fulfil the wider 4 role -- but what would have happen if the 5 initial call had been handled by a less 6 experienced SSC person? 7 "If there is a similar case in future, where 8 the system is being blamed, would it not be 9 sensible to have a technical review of all of 10 the evidence, at the first indication that 11 a case may be going to court? Someone involved 12 in that review would then be well placed to give 13 evidence in court." 14 Again, is that the same point but put in 15 a different way? 16 A. Yes, I think it probably is and I would say that 17 that, I think, is the role that Gareth Jenkins 18 then picked up. 19 Q. You refer there to a technical review of all of 20 the evidence. What did you have in mind? 21 A. Anything you could lay your hands on that might 22 be relevant. That was how SSC investigated. 23 But it's -- I mean, specifically, it would -- 24 the starting point would always be the message 25 store for the time, and anything else. 68 1 Q. On the point that you found yourself being 2 treated as an expert witness, we're of course 3 now aware that other Fujitsu employees attended 4 court and gave evidence against subpostmasters. 5 Were you aware of that at the time of writing 6 this memo? 7 A. At the time of writing this, I think I was only 8 aware that people in the Security team had 9 appeared in court and I'm not sure, I think 10 Gareth had perhaps expected to appear as 11 a witness before this point but, in fact, hadn't 12 done so, but he probably produced witness 13 statements. But I wasn't aware of anybody else 14 in Fujitsu -- 15 Q. After this time, did you come to know that 16 Mr Jenkins was giving evidence in various trials 17 around the country? 18 A. Yes. 19 Q. Did you discuss with him this issue that you 20 raise here, namely whether you were being 21 treated as an expert witness or a witness of 22 fact and whether that was a problem or a concern 23 for him? 24 A. I don't remember discussing it explicitly with 25 him. 69 1 Q. Do you know if this document was sent to him? 2 A. I don't know. 3 Q. Did you send it to him? 4 A. As I've said, I've got no recollection of -- 5 I had forgotten that I produced this document. 6 If he wasn't on the distribution list of the 7 email, then, no, I don't think I would have sent 8 it to him. I think I felt it was something -- 9 well, I think I sent it initially to my manager, 10 really, to send on to the Security team, and my 11 manager's management. 12 Q. So in the years that followed, you didn't 13 discuss with Mr Jenkins the status of the 14 evidence that he was going to give or had given 15 or his understanding of his status? 16 A. No, I don't think I ever discussed his specific 17 status. I think perhaps I assumed, because he 18 knew so much about everything, he was an expert 19 witness. But that's -- in the legal sense, 20 I wouldn't have known precisely what was meant 21 by that. 22 Q. Can we look at section 3, please, "Disclosure of 23 evidence". You say: 24 "Fujitsu made a major legal blunder by not 25 disclosing all the relevant evidence that was in 70 1 existence. I found myself in the invidious 2 position of being aware that some information 3 (Tivoli event logs) existed, but not sure 4 whether they had been disclosed or not, since 5 I had not been party to any of the requests for 6 disclosure. It became evident in court they had 7 not been disclosed. 8 "Quoting from an email received from [the 9 Post Office's] solicitor after my revelation 10 ..." 11 This was the email you referred to earlier 12 this morning? 13 A. Yes. 14 Q. This from Mr Dilley, we needn't track the email 15 down because you've cut and pasted it accurately 16 into this document: 17 "'In any litigation, the parties involved 18 have a continuing obligation pursuant to the 19 Court rules to disclose all documents that may 20 help or hinder their case or the other side's 21 case. In this context a 'document' means 22 anything in which information of any description 23 is recorded, so it includes, just for example, 24 a computer database. Previously, I had asked 25 Fujitsu to let me have all the info it had and 71 1 had been helpfully given HSH call logs, 2 transaction logs and events logs. I was 3 recently told that there was a message store 4 which had everything else on it and we invited 5 Mr Castleton to look at this, but he didn't take 6 up the opportunity.' 7 "This suggests that disclosure of the 8 message store itself was an afterthought, though 9 it is fundamental to the system. I know that 10 for fraud cases the 'transaction log' and 'event 11 log' are extracted from the full message store 12 and submitted, but surely the full message store 13 has to be disclosed in all cases?" 14 Just stopping there, you say, "I know that 15 for fraud cases ..." That tends to suggest that 16 you did know that there were another species of 17 case being conducted at this time? 18 A. Yes, I suppose so. I'm -- yes, I'm not sure 19 quite why I made that distinction, really, but 20 I knew that the ARQ data, which is what we're 21 talking about there, could be obtained and 22 I suppose I had assumed that was for -- 23 specifically for fraud cases, but ... 24 Q. You say that transaction log and event logs are 25 exacted and submitted in fraud cases. Who did 72 1 that? 2 A. That's the ARQ extract that was done by the 3 Security team. 4 Q. When you say "and submitted", do you mean and 5 submitted to the court? 6 A. To Post Office, is probably what I meant. 7 Q. You say: 8 "... but surely the full message store has 9 to be disclosed ..." 10 Is that because, as you've earlier said, 11 it's fundamental to the system? 12 A. Yes, and I think until I'd seen that email from 13 Stephen Dilley, I think perhaps I hadn't 14 realised that it wasn't disclosed initially. As 15 I said, I didn't know what had been disclosed 16 and what hadn't. But I was very surprised to 17 find that that seemed to be something he'd only 18 heard about recently. 19 Q. So would you agree that the full message store 20 from the branch had to be disclosed in all cases 21 on which reliance on Horizon data was made by 22 the Post Office and merely disclosing standard 23 filtered ARQ data didn't meet the disclosure 24 requirement that had been described to you in 25 this email? 73 1 A. It certainly didn't seem to meet the disclosure 2 requirement that was in this email. I'd only 3 had that email well into the, you know -- this 4 was Christmas 2006. You know, I was surprised. 5 Q. Were the Security department, to your knowledge, 6 aware of all of the files archived to audit 7 servers that held relevant material? 8 A. Yes, I didn't know what they were aware of, 9 I thought -- yes, I think that paragraph there 10 sort of sums up my knowledge. 11 Q. That's the next paragraph? 12 A. Yes. 13 Q. "Many other files are also archived to the audit 14 servers as a matter of course and could hold 15 relevant information, although the Security team 16 are not necessarily aware of their existence or 17 potential relevance. I'd like to suggest that 18 a list of these files is compiled so that 19 similar mistakes are not made in future." 20 Starting with "Many other files are also 21 archived to audit servers", what other files are 22 you speaking of? 23 A. All sorts of things. You know, it's an enormous 24 computer system, so a lot of back-end files, if 25 you like, were continually being generated. 74 1 Some of them would -- were transient, others 2 were backed up in one way or another. Some of 3 them were possibly written to the audit servers, 4 although SSC wouldn't have seen those because we 5 didn't have access to the audit servers. So 6 I didn't know what might be there but I had 7 a feeling there might be a lot of stuff. 8 This wouldn't be specifically counter files 9 because some of the diagnostic files that were 10 written that only existed on counters, they 11 wouldn't have been maintained in this way, but 12 there could have been files containing 13 transactions for a branch, as well as for lots 14 of other branches, that were still in existence. 15 I didn't -- it was really a bit of 16 an unknown and I thought, well, perhaps somebody 17 ought to try to make sure they know it in case 18 this is relevant in future. 19 Q. But a known unknown? 20 A. Yeah. 21 Q. You suggested a list of files being compiled 22 because the Security team might not be aware of 23 the existence of such files and therefore their 24 relevance? 25 A. Yeah. 75 1 Q. To your knowledge, was that done? 2 A. I never had any follow-up on any of the things 3 in this document. 4 Q. You continue: 5 "And what about calls on PEAK, which may 6 have evidence attached? And any evidence which 7 might have been kept within SSC? I was not 8 asked whether I had anything that might have 9 been relevant (as it happens, in this case I did 10 not)." 11 Is that because you hadn't attached anything 12 to the PEAK? 13 A. Yes, I didn't attach anything to the PEAK and 14 anything that I did have in file store in 2004, 15 I would have kept it for a year, year and a half 16 but then I'd have had a tidy-up and got rid of 17 it. 18 Q. But your point was there needs to be a more 19 systematic approach to this -- 20 A. Yes -- 21 Q. -- namely people being asked to give evidence -- 22 A. Yes. 23 Q. -- should be asked to turn over relevant 24 material? 25 A. Yes, I felt that that probably -- going on what 76 1 Stephen Dilley was saying should have been 2 disclosed, then surely that would have included 3 those sort of things. 4 Q. You continue: 5 "Of course there may be subtleties to this 6 that I am unaware of, whereby data may exist but 7 there is no obligation to disclose it. If this 8 is the case, could any future witnesses be 9 briefed appropriately? The response 'no one has 10 ever asked for that before' does not seem to be 11 a good reason for non-disclosure." 12 Who had given the response "No one has ever 13 asked for that before"? 14 A. I cannot now remember but, since I put it there, 15 it suggests that somebody may have said it. 16 Q. Within Fujitsu? 17 A. Yes, this was all aimed within Fujitsu. 18 Q. Helpdesk calls the last section, section 4: 19 "This case highlighted a common problem, 20 both in 2004 and now. The postmaster raised 21 many calls about his continuing losses, both 22 with Horizon and with the NBSC. These kept 23 being bounced and it took weeks before a call 24 was passed to SSC." 25 You're essentially there referring to the 77 1 evidence I took you through yesterday morning. 2 A. Yes. 3 Q. "Strictly speaking, problems with discrepancies 4 do need to be investigated by NBSC in the first 5 instance, but where there are continuing 6 unresolved problems it should be possible to get 7 the issue investigated properly, and one of the 8 Helpdesks should be prepared to take 9 responsibility for the incident. Personally 10 I think the fact that the Horizon helpdesk is 11 penalised for passing 'Advice and Guidance' type 12 calls on to third line leads to too many calls 13 being closed without proper investigation or 14 resolution. This is very frustrating for 15 postmasters, though possibly not an issue of 16 concern to [the Post Office]." 17 So, first of all, you say you think the fact 18 that the Horizon Helpdesk is penalised from 19 passing "Advice and Guidance" type calls to the 20 SSC? 21 A. Yes, there was some -- 22 Q. What was the penalty? 23 A. I cannot now remember whether it was just 24 a black mark or a financial thing. I've no idea 25 that -- 78 1 Q. What are you getting at here? What's the 2 underlying problem here? 3 A. They were reluctant sometimes to pass calls to 4 SSC if they thought they were just going to be 5 told off for having passed over a call that they 6 shouldn't have done. 7 Q. So, what, they bounced it back to the NBSC? 8 A. I think that did happen in some cases. It 9 may -- 10 Q. So what we saw in the many documents I took you 11 through yesterday morning was something that 12 wasn't isolated to this case; it was a recurring 13 problem? 14 A. Yes, it was. 15 Q. What we've seen is emblematic of a wider 16 problem? 17 A. I think so, yes. Certainly, the calls we looked 18 at yesterday, I feel that there was -- there 19 were -- there was at least one where I wished 20 they had passed it on to SSC, possibly two, and 21 it would have avoided some of the toing and 22 froing. I don't believe it would have made any 23 difference to the outcome but -- 24 Q. You say it leads to too many calls being closed 25 without proper investigation or resolution. Why 79 1 did it lead to calls being closed without proper 2 investigation? 3 A. Because the Helpdesk were just bouncing them 4 back, without it coming to SSC, and there 5 probably were cases where, you know, there was 6 a system error that did need to be picked up on. 7 And, obviously, in that case, the sooner it came 8 to SSC, the better. 9 Q. So that would just be bounced back and, if the 10 postmaster wasn't persistent -- 11 A. That did happen, yes. 12 Q. -- it would just we closed off? 13 A. Yes. That could -- that did happen. 14 Q. What would happen if there was a loss? They 15 would just have to pay up? 16 A. If they couldn't find the reason for the loss as 17 a business issue, as I say, discrepancies were 18 most likely not to be system problems but they 19 could be and, certainly we'll see when we go 20 through the rest of the bugs, errors and 21 defects, that there were cases where it should 22 have been reported to SSC years before it 23 actually was. 24 And there was this, you know -- we got calls 25 through to SSC that probably shouldn't have been 80 1 passed through to us but there were also others 2 that should have come to us that didn't reach 3 us. 4 Q. So all the while the system error was 5 continuing? 6 A. The one I'm thinking about yes, although -- 7 Q. Which one are you thinking about in particular? 8 A. The one that gave 14 branches a loss every 9 February. 10 Q. That went on for a number of years? 11 A. Two years. 12 Q. Yes. 13 A. Same branches. 14 Q. You say: 15 "This is very frustrating for 16 postmasters ..." 17 A. Yes, I think Mr Castleton would probably agree 18 with that. 19 Q. You're not basing that opinion just on 20 Mr Castleton's case? 21 A. No. 22 Q. That's, at the time of writing, your six or 23 seven years' accumulated experiences; is that 24 right? 25 A. Yeah, yeah. 81 1 Q. You say it's possibly not an issue of concern to 2 the Post Office. Why did you think the Post 3 Office was possibly not really concerned about 4 this? 5 A. I think, by this point, I had realised that the 6 outcome for the postmasters was not Post 7 Office's primary consideration. 8 Q. What was the Post Office's primary 9 consideration? 10 A. I think, by this point, it was fairly clear that 11 they were keen on defending the integrity of 12 their system rather than trying to get to the 13 bottom of issues, whether system problems or 14 business problems, affecting individual 15 branches. 16 Q. So it was more important, in your accumulated 17 experience, to the Post Office to defend the 18 integrity of Horizon, rather than conduct 19 a proper investigation to determine whether the 20 system is causing discrepancies? 21 A. Whether the system or business practices at the 22 Post Office are causing the discrepancies. 23 Q. Can you help us: that overarching view that 24 you've just expressed, on what information or 25 evidence was it based? 82 1 A. I think, partly in the case of Mr Castleton, 2 that they hadn't -- they didn't appear to have 3 made any attempt to help him get to the bottom 4 of the problems that he was having and that it 5 was just seen that "Oh, well, you know, he's 6 signed these things off, therefore that's his 7 responsibility". 8 When I started working on Horizon, I didn't 9 appreciate the fact that the subpostmasters 10 really didn't work for Post Office but they were 11 a third party in all this. Any other systems 12 I'd ever worked on, if there was a problem and 13 I said, "Well, I can't see anything wrong on the 14 system side", then the customer, if they 15 continued to have problems, would push back 16 quite strongly and say, "But look, this isn't 17 right, we need to sort it", and then we would 18 work together to sort it out. 19 But with the case of Post Office -- and it 20 took me quite a long time to realise how it 21 worked -- the postmasters had -- they didn't 22 have the power to do the pushing back and our 23 client, Post Office, who would have had that 24 power, did not seem interested in doing that. 25 Q. How was that manifested itself, outside of 83 1 Mr Castleton's case? 2 A. I just think in other cases where I had looked, 3 where there were discrepancies or problems, and 4 I was not able to find any system problem and 5 was pretty sure there was not a system problem, 6 but I could -- you know, they were having losses 7 or whatever, and I would say "Well, you know, 8 perhaps your manager can help you resolve this", 9 and the postmaster's view on that bit of advice 10 was not usually indicative that they thought 11 that that would be successful. 12 Q. This is you expressing that contemporaneously, 13 back in 2007, rather than now -- 14 A. Yeah. 15 Q. -- through the sentence "possibly not an issue 16 of concern to the Post Office"? 17 A. Yeah. 18 Q. In that sentence? 19 A. Yeah. 20 Q. Standing back, do you agree that you were, in 21 this afterthoughts document, raising a series of 22 fundamental and important issues about the 23 process of giving evidence in court, in 24 proceedings which concerned data produced by the 25 Horizon System? 84 1 A. Yes, I mean, I was doing it really from my 2 personal point of view, having been through this 3 process. I just felt it needed to be fed back 4 as things that concerned me. 5 Q. You raised an issue about the need for a proper 6 technical review of a wide range of data before 7 proceedings were even launched, yes? 8 A. That was a suggestion. 9 Q. You raised an issue over confusion as to whether 10 a witness was giving evidence of fact or opinion 11 evidence? 12 A. Yes. Yes, I don't think anybody had ever 13 mentioned opinion evidence as an option. 14 Q. You raised an issue about witnesses being asked 15 to speak about the reliability of Horizon more 16 generally, rather than about the narrow work 17 that they had actually done? 18 A. I don't think that was particularly a concern. 19 It was just being -- it was just sort of being 20 asked not necessarily about the reliability but 21 about anything that was outside what I thought 22 I was meant to be talking about. 23 Q. These are all issues you now know, I think, 24 which have come to afflict the presentation of 25 the Post Office's cases against subpostmasters 85 1 in criminal proceedings? 2 A. Yes. 3 Q. You were then, back in 2007, describing issues 4 that may have afflicted past and then current 5 criminal prosecutions, albeit you had no 6 knowledge of those? 7 A. Yes. 8 Q. Would you agree that you were raising a series 9 of red flags? 10 A. Yes, I'm not sure I thought about that at the 11 time -- thought about it like that at the time. 12 I just thought there were lessons to be learnt. 13 Q. Would you agree that it was important for both 14 Fujitsu and the Post Office to address these 15 issues and to address them properly and 16 promptly? 17 A. I was pretty clear on my point of view and, yes, 18 I hoped it might have some impact for the 19 future. 20 Q. You weren't raising them to be ignored -- 21 A. No, no. 22 Q. -- nor as an insurance policy against what you 23 had done? 24 A. No, not at all. 25 Q. You weren't going to give evidence again, to the 86 1 best of your knowledge -- 2 A. That's true, yes. 3 Q. -- and this was to help others, presumably -- 4 A. Yes. 5 Q. -- whether those others be subpostmasters or 6 your colleagues? 7 A. Yeah. 8 Q. Can we see what was done with your report 9 please, and look at FUJ00152300. Can we see, at 10 the foot of the page, please, an email of 11 29 January 2007 -- that's the date of your 12 report, remember, 29 January 2007 -- 13 A. Yeah. 14 Q. -- from Mik Peach to Brian Pinder, Security 15 Manager; is that right? 16 A. Yes. 17 Q. Naomi Elliott, that is Mr Peach's manager; is 18 that right? 19 A. That's Mr Peach's manager, yes. 20 Q. And copied to you? 21 A. Yes. 22 Q. "'Mop up' on the Castleton case", subject: 23 "Brian, 24 "I understand from Anne that you do not 25 intend to have an internal review on the 87 1 Castleton case." 2 Stopping there, had you asked Mr Pinder 3 whether there was going to be an internal 4 review? 5 A. I have no memory of that but I assume I had 6 done. 7 Q. That tends to suggest you had -- 8 A. Yes. 9 Q. -- and he'd said no? 10 A. Yeah, I assume so. 11 Q. "Nevertheless, we are concerned that POA ..." 12 That means Post Office Account? 13 A. Yes. 14 Q. Does that mean Fujitsu? 15 A. Yes, Fujitsu. 16 Q. So the Fujitsu Post Office Account: 17 "... made some errors during the course of 18 this case which could prove critical in any 19 future litigation. 20 "To this end, Anne has written up her 21 thoughts and comments (attached), and I would 22 welcome your comments." 23 Your document is an attachment. 24 A. Yes. 25 Q. If we scroll up and see what the reply was, 88 1 please. 5 February, so about a week later -- 2 A. Yes. 3 Q. -- an email addressed to you and Mr Peach, 4 copied to Naomi Elliott. Then in the title 5 addressed to you, "Mik, Anne": 6 "Thanks Mik, there was no intention to have 7 a wash up on this particular case as such but 8 I must stress that from the outset this was 'new 9 ground' and a particularly unusual case (1st of 10 its kind in 10 years) for all concerned. It 11 involved many different variables which, at any 12 point in time could have culminated in a totally 13 different outcome. 14 "This enquiry took well over a year to 15 conclude and routine procedures which have 16 served us well for 10 years were suddenly being 17 stretched to new limits, but it does highlight 18 how (POA) can be called to account and I totally 19 agree we must learn from this. 20 "Anne (many thanks for your comments) you 21 have highlighted some interesting areas of 22 procedure which we need to recognise, and I will 23 custody these with Naomi and keep you both 24 informed." 25 Did you ever hear anything again? 89 1 A. Not that I recall but, since I'd forgotten this, 2 who knows. 3 Q. We haven't got a record of anything else 4 happening -- 5 A. No. 6 Q. -- as a result of this? 7 A. No. 8 Q. Would it be unfair to describe this as a pat on 9 the head? 10 A. Yes. No, sorry, not unfair. It would be fair. 11 Q. It's fair. "Well done, Anne, thanks"? 12 A. Yes. 13 Q. "We're just filing this"? 14 A. Yes. 15 Q. Is that how you read it? 16 A. Yes. 17 Q. Overall, your view was that something was going 18 wrong at Marine Drive, you couldn't see what the 19 problem was and the only way to progress matters 20 was at the branch? 21 A. Yes. 22 Q. That needed Post Office to take some action, 23 didn't it? 24 A. Yes. 25 Q. So far as you were aware, that wasn't done? 90 1 A. That didn't appear to be done. 2 MR BEER: Thank you very much. Those are the only 3 questions I'm going to ask on the Castleton 4 case. Thank you. 5 A. Okay. 6 MR BEER: Sir, we're going to move now to look at 7 some of the bugs, errors and defects held over 8 from last time. I wonder whether I could 9 impertinently ask for a lunch break now, because 10 it's a useful stopping off point, and break 11 until 1.30? 12 SIR WYN WILLIAMS: Of course, Mr Beer. So we'll 13 resume at 1.30. 14 MR BEER: Thank you very much, sir. 15 (12.25 pm) 16 (The Short Adjournment) 17 (1.30pm) 18 MR BEER: Sir, good afternoon, can you see and hear 19 me? 20 SIR WYN WILLIAMS: I can, thank you. 21 MR BEER: Thank you very much. 22 Good afternoon, Mrs Chambers. Before we 23 turn to look at, in fact, just one of the bugs, 24 may we just return to a question I asked you 25 before lunch. Do you remember your 91 1 afterthoughts document? 2 A. Yes. 3 Q. I asked you whether you had shared that with 4 Mr Jenkins? 5 A. Yes. 6 Q. You said, unless the email trail showed that you 7 had, you couldn't recall having done so; I think 8 that's a summary of your evidence. We looked at 9 the email distribution in the email I showed you 10 right before lunch. 11 Never mind showing him the document or 12 sending him the document, do you recall, after 13 writing it, discussing any of the four issues 14 with him? 15 A. I certainly don't remember any formal 16 discussion. Whether we had an informal chat, 17 I have no recollection of that. I can't say 18 "yes" or "no" on that. 19 Q. Just breaking it down, a conversation or 20 conversations with him about the need to conduct 21 a fundamental review of data before a legal case 22 was commenced? 23 A. So we may -- I mean, I think he might have been 24 of the same view of that anyway, whether we 25 actually had a discussion along those lines as 92 1 to how necessary it was, I can't be certain. 2 I do know that subsequently when he was 3 preparing for other court cases, he did do -- he 4 appeared to be doing a fairly thorough 5 examination because occasionally he'd ask me to 6 double check some of the things he was looking 7 at as well, but I'm not sure -- I don't think 8 this was ever formally documented anywhere. 9 Q. Was that -- when he was doing what appeared to 10 be a thorough examination -- at the stage at 11 which he was preparing to be a witness giving 12 evidence or was it the issue that you were 13 concerned with, which is conducting 14 a fundamental review of data before proceedings 15 are even commenced? 16 A. I don't think he would have been involved until 17 proceedings were commenced. 18 Q. What about the other issue that you raised in 19 the paper concerning giving evidence of fact or 20 opinion evidence as a witness? 21 A. I don't think I ever discussed that with him. 22 Q. Did you ever discuss with him the question of 23 creating a list of available data for provision 24 to the Security team, so they would know what 25 was there in order that disclosure obligations 93 1 might be complied with? 2 A. I've no recollection of discussing that with him 3 and I don't think he'd have been in a position 4 to have made that list anyway. 5 Q. Who was responsible for taking forward the four 6 issues that you had raised in your paper? 7 A. I don't know. I was concerned that nobody 8 seemed to be doing any sort of a follow-up so 9 I sent it to my manager. So, at that level, 10 I was really passing it on to him and, as we'd 11 seen, he then passed it up to both his manager 12 and the Security team manager. 13 Q. Thank you. 14 Can we turn to bugs, errors or defects. In 15 the light of the approach that the Inquiry, the 16 Chairman, is taking to Mr Justice Fraser's 17 judgments and the significant quantity of 18 material that we've now got in relation to bugs, 19 errors and defects and the helpful explanations 20 you've given in your first witness statement in 21 relation to some of them, I'm just going to 22 concentrate, if I may, on the suspense account 23 bug. 24 A. Right, which number is -- 25 Q. It's 3, bug 3. 94 1 A. Oh, okay. Yeah. 2 Q. Would you agree with the following summary of 3 the suspense account bug: firstly, it's 4 a Horizon Online bug? 5 A. Yes. 6 Q. Secondly, it involved branches that deleted 7 a stock unit at the end of 2010 with 8 a transaction in local suspense. They were 9 affected as a result of that change? 10 A. Yes. 11 Q. Thirdly, in essence, the transaction that was in 12 the local suspense of the deleted stock unit was 13 left in the database used to construct branch 14 trading statements in the same trading period in 15 the following years? 16 A. Yes. 17 Q. Fourthly, this would cause a false discrepancy. 18 It caused discrepancies in 2012 but they weren't 19 identified as a bug at that stage? 20 A. Yes. 21 Q. The bug was only discovered, lastly, in February 22 2013? 23 A. Yes. 24 Q. Okay. Can we look, then, and pick the story up 25 with one of the initiating PEAKs of February 95 1 2013, FUJ00081875. Can you see this is a PEAK 2 numbered PC0223870? 3 A. Yes. 4 Q. The summary is: 5 "Branch [and then a FAD code is given] has 6 an unexplained discrepancy"? 7 A. Yes. 8 Q. If we just scroll down, please. We can see that 9 it's opened on 25 February 2013? 10 A. Yes, sorry, yeah. 11 Q. Yes? Earlier on, you referred to a bug that it 12 had been slow to recognise that had affected 14 13 branches? 14 A. Yes. 15 Q. Is this the bug you are referring to? 16 A. This is the bug that I was referring to, yes. 17 Q. I think you said at the time that that was in 18 part because of a failure at the NBSC and 19 Helpdesk area to escalate to the SSC earlier? 20 A. Yes, I can't remember now whether this -- 21 I don't think this actually got as far as the 22 Horizon Helpdesk. I'm sorry, I can't remember 23 now. I think there were calls at NBSC, it was 24 discovered subsequently, when checks were made, 25 and I can't remember if none of them were passed 96 1 on to the Horizon Helpdesk or just one but 2 nothing got as far as SSC until 2013. 3 Q. So, essentially, there's a bug in the system 4 causing discrepancies -- 5 A. Yes. 6 Q. -- that has been reported by a postmaster but 7 has not been escalated to the SSC? 8 A. Yes. 9 Q. Therefore, the bug continues to work the 10 discrepancies. 11 A. Yes. 12 Q. So the summary of the call, the branch has 13 I think that is, an unexplained discrepancy, 14 yes? If we then scroll down, please, further 15 notes: 16 "The Branch has an unexplained discrepancy. 17 They balanced and rolled trading period on 18 6 February, they have one stock unit, AA, and 19 this was balanced with a loss of £39.57, which 20 was transferred to local suspense however the 21 figure that was cleared out from local suspense 22 was much higher £9,839.45." 23 A. Yes. 24 Q. "I have carried out transaction logs for all 25 transactions from date range 31 January to 97 1 6 February. Branch has submitted copies of the 2 final balance reports from [Trading Period] 3 10BP4 ..." 4 What does that mean? 5 A. Yes, by this time instead of having weekly cash 6 account periods we had four-week trading periods 7 and they could be split into weekly balance 8 periods if postmasters wanted to balance things 9 weekly. 10 Q. So this is balancing period 4 of the trading 11 period 10 -- 12 A. Yeah. 13 Q. -- and balance period 5 of trading period 10; 14 yes? 15 A. Yes. 16 Q. "The balance report for TP10 BP5 shows 17 discrepancy transferred of £39.57 and then 18 discrepancy resolved of £9,839.45. The 19 transaction log completed for all transactions 20 does not show any other figures being entered 21 into or removed from the housekeeping/local 22 suspense account." 23 A. Yes. 24 Q. So this text here, is this taken from the NBSC? 25 Is this an NBSC person writing it? 98 1 A. Yes, it appears so. The contact was Ibrahim at 2 NBSC, and he had probably sent this to HSH, 3 whatever they were called then, quite possibly 4 as an email or something. 5 Q. So it's been cut into the PEAK? 6 A. It's been put into a new PowerHelp call, or 7 whatever calls were called at that point, which 8 has then be routed to PEAK and has created 9 a PEAK in the process. 10 Q. Got it. Then if we scroll down, please. There 11 is some process chat, yes? 12 A. Yeah. 13 Q. You changed the summary -- 14 A. Yeah. 15 Q. -- from "Branch has an unexplained discrepancy" 16 for putting the branch code having 17 an unexplained discrepancy? 18 A. That was standard practice. We always tried to 19 put the branch code in the heading of the call. 20 Q. Is that for subsequent searching purposes? 21 A. It would have been picked up for searching 22 wherever it was but just so -- it just made it 23 clearer and then, if you were looking at the 24 stack, you might possibly see if you've got 25 several -- 99 1 Q. So then the call having been assigned to you at 2 5.02, 45 minutes later you make the change and 3 then, the following day, you make your first 4 substantive entry; is that right? 5 A. Yes. 6 Q. If we scroll down so we can read the whole of 7 that: 8 "When they completed the balance on the 9 6 February and cleared the loss from local 10 suspense, the amount cleared was £9,000-odd, 11 instead of the loss they put into local suspense 12 which was £39-odd. This appears to be 13 a consequence of something that happened during 14 the previous trading period rollover on 15 2 January: a 'gain to local suspense' of 16 £9,000-odd was included in the DEF opening 17 figures." 18 What are the "DEF opening figures"? 19 A. "DEF" was a default stock unit, I can't now 20 fully remember its purpose but every branch, by 21 this point, had one of those. 22 Q. "I don't think any of the local suspense 23 products should ever appear in the opening 24 figures." 25 Can you explain what you mean there, please? 100 1 A. Because when you rolled over into a new trading 2 period, before you did that, you had to clear 3 local suspense, ie make good any loss or gain 4 that had been made during that period or clear 5 it in some other way, and so local suspense 6 should always be zero at the point that you roll 7 over, so there should never be any value for it 8 in the opening figures for the next period. 9 Q. You continue: 10 "However I found 14 such lines, all product 11 6295 gain to LS." 12 What does that mean. 13 A. Gain to local suspense. 14 Q. What does "product 6295" mean? 15 A. Every product had a number as well as a name and 16 the number for gain to local suspense was 6295. 17 Q. "Unfortunately all created November to December 18 last year so there is almost no remaining 19 counter evidence." 20 What did you mean by that? 21 A. Because these records had been created some 22 three months -- two/three months previously, 23 looking at log files on the counter, and so on, 24 was unlikely to help. 25 Q. How did you find the 14 other examples -- 101 1 A. Um -- 2 Q. -- or instances? 3 A. By this point, we're not dealing with Riposte 4 and message stores, and so on. 5 Q. No. 6 A. We're dealing with an Oracle database system 7 that was all held centrally, something called 8 the Branch Database. It should really have been 9 the Branches' Database because it contained -- 10 there were a lot of different tables containing 11 data for all the branches, obviously each record 12 said which branch it referred to, and so on. 13 So once I'd found the table of interest, 14 I could just do a query on that table to find 15 all records for that product in that table, 16 regardless of branch. 17 Q. You're continuing to investigate the cause and 18 implications? 19 A. Yes. 20 Q. Mr McEwan changes the call priority? 21 A. Yes. 22 Q. Is that increasing its priority? 23 A. That's increasing the priority. He was -- the 24 team leaders within SSC took it in turns to be 25 sort of duty manager for a day or a week or 102 1 something. He must have been the duty manager 2 on this occasion and I probably discussed it 3 with him and said, "This looks serious". 4 Q. You then include -- I think that's essentially 5 a hyperlink, I would call it, to BRSS Extracts. 6 What are BRSS Extracts? 7 A. The BRSS was a copy of the branch database, 8 which in fact was what we ran our support jobs 9 on, so we weren't impacting the live database. 10 Q. You were doing that presumably so you've got 11 a document to work from easily obtainable, the 12 reason -- 13 A. So this -- 14 Q. -- you created the hyperlink? 15 A. Yes, this was evidence that I'd added. It was 16 probably an Excel spreadsheet. 17 Q. You say: 18 "We only keep opening figures for three old 19 trading periods so can't be sure when the 20 problem started." 21 Does that mean that the opening figures for 22 three four-week periods were that which was 23 retained. 24 A. Yes, I believe so because old data was 25 continually having to be deleted out of the 103 1 tables. 2 Q. Continuing later that day, if we scroll down, 3 please, you ask Ibrahim at the NBSC if the 4 suspense report from 2 January is available. 5 He'll obtain it and email it to HSD, IMD -- is 6 that the Incident Management Team. 7 A. Yes, I believe so. 8 Q. What did you want that for? 9 A. I thought it would be interesting to see what 10 entries were on the suspense report. This was 11 a report that would have been produced at the 12 branch on 2 January. 13 Q. You got that by 4.51. You looked at it and it 14 didn't show an anomaly; is that right? 15 A. I couldn't see anything on it, yeah. 16 Q. Scroll down, please. You say you have asked, 17 top of the page "what the branch did about the 18 problem last year". Who were you asking there? 19 A. I can't remember whether I asked Ibrahim or 20 whether, by this time, I was talking direct to 21 the branch. I think it's more likely I asked 22 Ibrahim. 23 Q. By 4.20, you say you found the cause of the 24 problem. 25 A. Yes. 104 1 Q. "... some data from autumn 2010 has been 2 retained in [a table] for 14 separate branches. 3 These branches will all have been affected by 4 this problem late 2011 and late 2012, though in 5 some cases the amounts involved are small." 6 Do you then set out the 14 affected 7 branches? 8 A. Yes, I do. 9 Q. You wanted, however, to see what needed still to 10 be ascertained and you, I think, then set out 11 a series of questions and answers to yourself? 12 A. Yes, I mean, any investigation you've got 13 various things that then need to be done and 14 this was trying to record what needed to be 15 done. 16 Q. Do we see those at the foot of the page and 17 continuing? 18 A. Yes. 19 Q. So under "Still to be investigated"? 20 A. Yes, as far as I can see. 21 Q. Thank you: 22 "Exactly how did these records cause the 23 observed effect? 24 "Why were these records not removed by the 25 normal archiving process? 105 1 "What impact has the problem that on the 2 branch accounts? 3 "What impact has the problem had on POLSAP? 4 "How do we remove the records to prevent 5 future problems? 6 "Were there any affect branches which have 7 since closed? 8 "There are 19 other branches which have had 9 old data in the affected table, but not relating 10 to Local Suspense. Could this cause any problem 11 with the branch accounts?" 12 What did you mean by that last entry, 13 please? 14 A. When I looked in the table that I had 15 identified, that had these 14 local suspense 16 lines in, I found that there was other old data. 17 I, presumably at this point, just looked for any 18 records older than a number of months because 19 they should have been removed by the archiving 20 process. I found that there was some other data 21 in there but, as I say, not relating to local 22 suspense, so I was -- obviously that was 23 something else that needed to be followed up. 24 Q. The next day, I think you repeat your questions 25 and then give some of the answers to them; is 106 1 that right? 2 A. Yes. 3 Q. So we're now on the 28th: 4 "Exactly how did these records cause the 5 observed effect?" 6 Then you describe essentially the problem 7 that I outlined at the beginning, yes? 8 A. Yes. 9 Q. "How to identify the problem from branch 10 reports? 11 "Branch trading statement: the sum of two 12 Discrepancy Transferred lines does not match the 13 total of the two Discrepancy Resolved lines." 14 A. Yes. 15 Q. Can you explain how that identifies the problem 16 from branch reports? 17 A. Well, the branch trading statements, which -- it 18 was the cash account replacement, that would 19 show the discrepancies being put into local 20 suspense from all the different stock units and 21 then, before they could roll over into the next 22 trading period, the final -- when they rolled 23 over the final stock unit, then they had to 24 clear that amount that was in local suspense. 25 This particular problem that -- the whole 107 1 problem was because they were having to clear 2 more than they had put into local suspense, so 3 those numbers would not match up on the branch 4 trading statement. 5 Q. Thank you. Then if we scroll down, please, to 6 16.05.10, on 6 March. Thank you. You record 7 that: 8 "There was a conference call with [the Post 9 Office] (Laura Darby, Mark Wardle and others) on 10 28 February about this call, and the spreadsheet 11 showing the impact of the problem on the 14 12 branches were sent to them by Steve Bansal. We 13 are waiting to hear from Mark whether this is 14 sufficient information for them to resolve the 15 consequences on the branches and POLSAP." 16 A. Yes. 17 Q. What were the consequences on the branches? 18 A. They had had to clear a loss or a gain, which 19 had been -- was not the amount that they should 20 have had to clear. 21 Q. You say: 22 "We will then need to get the old data 23 causing the problem removed from the database 24 ..." 25 A. Yes. 108 1 Q. "... and consider whether extra checks should be 2 put in place to trap similar anomalies in the 3 future." 4 A. Yes. 5 Q. Were extra checks put in place to trap similar 6 anomalies in the future? 7 A. Yes, they were, eventually. 8 Q. When were they put in place? 9 A. I think it may be at the bottom of -- it's at 10 the bottom of one of the PEAKs. It wasn't for 11 several months, I think. 12 Q. What was the check put in place? 13 A. I believe there were two checks. One was as 14 I've already outlined and then there was 15 a second one, which I can't remember without -- 16 I'm sure I've written it down somewhere or it's 17 in the -- it's certainly in one of the PEAKs. 18 But this was -- you know, we knew we had 19 followed up these specific ones but it was just 20 in case such a thing -- situation should ever 21 occur in future. 22 There was no reason to think it would but, 23 if it did, then people would be alerted to it 24 rather than the postmaster having to raise 25 a call. 109 1 Q. Can we look, please, at POL00028744. Thank you. 2 This is, I think, a linked PEAK. 3 A. Yes, it's a clone of the one that we were 4 looking at previously. 5 Q. If we go forwards, I think, to 13 March, 6 please -- if we keep scrolling, please -- you 7 make an entry on this copy of the cloned PEAK: 8 "We need to make sure a copy of the records 9 being removed is kept somewhere in case there 10 are questions to be answered about any of the 11 affected branches' accounts." 12 Yes? 13 A. Yes. 14 Q. Firstly, the records being removed. What 15 records were being removed? 16 A. These were the records in this table, 17 BRDB_RX_BTS_DATA that had been created in 18 2010/2011 -- 19 Q. 2010. 20 A. -- which should have been removed by the 21 automatic archiving process but had not been 22 removed. 23 Q. You wanted a copy of that which was removed 24 retained? 25 A. It seemed like a good idea. 110 1 Q. You say "in case there are questions to be 2 answered about any of the affected branches' 3 accounts", was that in recognition that changes 4 to data that were being made like this might 5 lead to disputes between the Post Office and 6 subpostmasters, if there wasn't a clear and 7 defensible audit trail? 8 A. I don't think it was specifically that, it was 9 just obviously that there had been an error 10 caused at these branches and it just seemed 11 like -- we weren't deleting transaction data 12 here, this -- yes, they are financial records 13 but they shouldn't have been there but I just 14 felt we shouldn't just get rid of them without 15 making a note. 16 Q. You say "in case there are questions to be 17 answered"; questions to be asked by who? 18 A. Anybody. 19 Q. Including subpostmasters? 20 A. Yes, potentially, yes. 21 Q. So there did need to be a clear and defensible 22 audit trail of what had been removed in case 23 a subpostmaster asked about it? 24 A. Yes, I think so. 25 Q. Were subpostmasters told about the changes to 111 1 the data made by Fujitsu? 2 A. I can't remember now. I think Post Office dealt 3 with talking to the affected branches. As 4 I say, this was not changing transaction data. 5 Q. It was changing data that had financial 6 consequences for the subpostmasters? 7 A. In that it was data that shouldn't have been 8 there and so we were removing it. 9 Q. It was causing them to show a discrepancy that 10 was false? 11 A. Yes. Because these records, which should have 12 been removed, were still there. So we needed 13 a copy, so we could say, look -- if necessary we 14 could say "Yes, it was this record for your 15 branch and it caused a false discrepancy of this 16 amount". And we'd already got that information 17 into spreadsheets, and so on. I just felt it 18 was something that should be done. 19 Q. Do you recall whether the Post Office briefed 20 subpostmasters on the issue that had afflicted 21 their data, the steps taken to make corrections 22 and identifying what had occurred, so that any 23 further problems might be more easily spotted by 24 them when the known issue was impacting them? 25 A. I can't remember the details. I know we had 112 1 a couple of conference calls with Post Office 2 and I know Gareth wrote a note to attempt to 3 describe the problem, and there were discussions 4 between Gareth and Post Office as to what should 5 be done but, at this precise moment, I know 6 we've got documents about this, but I can't 7 remember the details. 8 Q. I'll try and help you. If we can look, please, 9 at POL00098189. If we start at page 4, please. 10 We can see from the foot of the page, this is 11 an email signed off by you. 12 A. Yes. 13 Q. Yes? 14 A. Yes. 15 Q. Then if we just go up to the bottom of page 3, 16 we can see it's an email sent by you to Andrew 17 Winn, Helen Love and the duty manager? 18 A. Yes. 19 Q. Who was Andrew Winn? 20 A. A Post Office person. 21 Q. Was it usual for you to communicate directly to 22 the Post Office in this way? 23 A. Not unless there'd been some previous conference 24 call or request. I wasn't sending -- as it 25 says, "as requested", and I believe this was 113 1 following one of the conference calls where 2 I had tried to explain this incredibly 3 complicated set of events which had led to this 4 problem, and I think everybody felt it would be 5 clearer if I tried to write it down. 6 Q. Thank you. If we scroll up, please. We can see 7 Mr Winn's reply. 8 "Hi Anne, 9 "Thanks for this. I've taken a deep breath 10 and tried to confirm the actual impact on 11 branches and prepare our communications to them. 12 Can you have a look at my first draft to see if 13 I have simplified and condensed what happened 14 without losing meaning or any key detail? 15 Appreciate you will not be comment on what the 16 branch or the FSC has subsequently done." 17 Some stuff about the Crown branches and then 18 an individual branch: 19 "Any thoughts? 20 "Andy." 21 Then scroll up, please. Keep going so we 22 can see all of your email. You deal with the 23 spreadsheet for Willen, which I think is one of 24 the Crown branches, and Lower Regent Street. 25 A. Yes. 114 1 Q. Then you say: 2 "I've made a few possible changes to both 3 letters which hopefully make the cause and scope 4 of the problem clearer. 5 "I tried to phone you, because Gareth 6 Jenkins mentioned that this problem is still 7 being discussed at a high level as part of the 8 ongoing investigations/checks into Horizon, and 9 I would hate anything I have put here to 10 compromise that -- I don't see why it should but 11 just wanted to flag it. I assume anything going 12 out to the branches will be reviewed in the 13 light of that." 14 A. Yes. 15 Q. So, essentially, he, Mr Winn, was asking you to 16 check his communication out to branches, his 17 draft communication out to branches for 18 accuracy; is that right? 19 A. For technical accuracy, yes. 20 Q. Yes, and was that usual practice for the Post 21 Office, to ask Fujitsu for advice on how it 22 should communicate with subpostmasters about 23 a problem arising in Horizon and impacting on 24 their Post Office business? 25 A. I don't recall it ever happening any other time 115 1 and I wouldn't say they're asking me how to 2 communicate. It was just trying to make sure 3 that any technical information in those letters 4 was as accurate as possible because, obviously, 5 you know, they'd be trying to make it clear to 6 the postmasters. 7 Q. In your first witness statement, you'd said 8 that, generally, when Fujitsu identified 9 a problem and communicated it to the Post 10 Office, it was for the Post Office to determine 11 if, how and when the problem should be 12 communicated to subpostmasters; is that right? 13 A. Yes. 14 Q. So is this something of an exception, ie you 15 involving -- or you being involved? 16 A. Yes, I think Post Office had made the decision 17 as to what they wanted to do and were then 18 looking to work out how to explain it clearly to 19 the postmasters, and it was just the technical 20 description of the problem as it appeared to the 21 postmasters. They presumably just wanted to try 22 to make sure that was accurate. But I am not 23 sure I should have been doing that and I was 24 obviously a little concerned that I was doing 25 it, but I was trying to be helpful. 116 1 Q. In the paragraph beginning "Tried to phone you", 2 you refer to the problem being discussed at 3 a high level. 4 A. Yes. 5 Q. Can you recall what the high level discussion 6 was about? 7 A. Presumably that there was a bug in Horizon. 8 Q. Yes, but why might a discussion at a high level 9 be compromised by what you had written or 10 approved to be written? 11 A. I don't know, but I think I was aware that I'm 12 not always very sort of politically sensitive, 13 and I was not wanting to get it wrong. 14 Q. So you didn't want to tread on toes; is that 15 right? 16 A. Yeah. 17 Q. So did you think you might be compromising the 18 position of Fujitsu in some way? 19 A. No, I don't think so. I just wanted to caveat 20 the fact that I was doing things and I didn't 21 want it just going out because I and Andy Winn 22 had said so. 23 Q. If we scroll up, please, to see Mr Winn's reply. 24 The first three paragraphs don't really matter 25 to us but then he says to you: 117 1 "I intend this to go through our solicitors 2 before it gets sent out. 3 "Have a good weekend." 4 So, presumably, at this stage, the issue was 5 being treated with sufficient seriousness by the 6 Post Office, that its communications out to its 7 subpostmasters were being vetted or approved or 8 at least looked at by their solicitors. 9 A. That's what that says. 10 Q. Yes. Was that usual? 11 A. I don't know. As I said, I was -- I don't 12 recall ever having been involved at this 13 particular level before. 14 Q. So this, from your perspective, is a bit of 15 an outlier, this type of thing that you were now 16 engaged in -- 17 A. Oh, yes. Yeah. 18 Q. -- ie communications between you and Mr Winn at 19 a working level, those higher up the chain 20 discussing at a management level perhaps 21 responsibility for the bug, and solicitors 22 checking communications to subpostmasters? 23 A. Yes, this was very unusual and my input was 24 really because it was such a complicated problem 25 to explain that, you know, people were still 118 1 coming back to me to try to, you know, 2 understand exactly which combination of rather 3 a lot of different things had caused these 4 branches to be impacted. 5 MR BEER: Thank you very much. They're the only 6 questions that I ask you about this bug and 7 they're the end of my questions. Thank you very 8 much, Mrs Chambers. 9 Sir, I think there are questions from one 10 Core Participant, the Hodge Jones & Allen team. 11 SIR WYN WILLIAMS: Yes. 12 MR BEER: Thank you. 13 Questioned by MS PAGE 14 MS PAGE: Thank you, sir. 15 Mrs Chambers, as you know I think, I appear 16 for number of subpostmasters including 17 Mr Castleton, who sits to my right. 18 A. Yes. 19 Q. Earlier on in your evidence, you were looking at 20 the ARQ data for week 41, which shows a zero 21 stamp declaration -- 22 A. Yes. 23 Q. -- and you explained that the stamp declaration 24 was one part of what went into the total stamps 25 carried forward into week 42 -- 119 1 A. Yes. 2 Q. -- and, in effect, it was the stamps bearing 3 denominations which come in large books, those 4 sort of stamps that say 1 penny -- 5 A. (The witness nodded) 6 Q. -- or they might say 10 pence, or whatever? 7 A. Yes, yes. 8 Q. Obviously that's the staple of any Post Office, 9 selling that sort of stamp? 10 A. Yeah. 11 Q. So we can probably agree that a subpostmaster 12 wouldn't accurately declare that to be zero. 13 You'd never get to the end of your stamps, would 14 you? 15 A. No, no. 16 Q. We can go to it, if you like, but you may be 17 able to take it from me that, in week 41, the 18 final balance, which has the breakdown of the 19 two parts, shows a figure for both parts? 20 A. Yes. 21 Q. So there isn't a zero. 22 A. Yes. 23 Q. There's stamps and there's the stock stamps, if 24 you like -- 25 A. Yes. 120 1 Q. -- the other stamps. That zero declaration that 2 we see before the cash account was finalised, 3 that appears to be an anomaly; is that fair? 4 A. It appears to be an anomaly, yes. But I do have 5 a probable explanation for it but not one that 6 I can prove. 7 Q. Well, I suspect that we probably don't need 8 probable explanations, if there's one that you 9 can prove then we could hear it, but -- 10 A. I don't have -- I mean, I do not now have any 11 evidence that I've seen that means that I can 12 prove it, but if I could explain what the 13 probable one is? Because we know that the 14 earlier declaration for £1,183 was included in 15 the accounts, so that zero declaration appears 16 to have had no effect because you would have 17 expected that to have wiped out that other 18 amount. 19 But when you made a stamp declaration or 20 a cash declaration, you could give a specific 21 drawer ID. I think we've seen that somewhere. 22 So, normally, Mr Castleton used drawer 11. Now 23 it is conceivable, but this is what I can't 24 prove, that this zero declaration had 25 a different drawer ID, which is why it didn't 121 1 overwrite the earlier one. But that, to me, 2 seems likely, because, otherwise, if it had been 3 the same drawer ID, it would have overwritten 4 the earlier one. 5 If I'd got the full message store that was 6 audited then that's something I could have 7 checked and I may well have checked it back in 8 2006 but I can't check it now, I'm afraid. 9 Q. No. All right. Well, we might come back to the 10 full extract from the message store in a little 11 while. On 2 February 2004, the second terminal 12 in Marine Drive, node 2, was the one that 13 failed, yes -- 14 A. Yes. 15 Q. -- and that was at around 14.20 -- sorry, it 16 came back into action around 14.20 on 17 2 February, having failed, yes? 18 A. Yes. 19 Q. If we could look at your witness statement, 20 please. If you need the reference it's 21 WITN00170200. If we go, please, to page 5, and 22 we go down to paragraph 17, please. 23 You deal with this issue of node 2 failing 24 and, in paragraph 17, having sort of dealt with 25 it and, if we go over the page, at the end of 122 1 the paragraph -- I'm so sorry, it's actually at 2 the end of paragraph 18, having dealt with it, 3 you say: 4 "In my view it is extremely unlikely 5 (bordering [on] impossible) that any 6 transactions were lost as a result of the 7 counter replacement." 8 Then you give your explanation of that: 9 "There is no indication of any problem with 10 the physical connection. Even if there were, 11 they could only impact CAP45 and not cause 12 losses in consequent periods." 13 So you've asked and answered your own 14 question: could the loss of node 2 or the 15 failure of node 2 have lost transactions? You 16 asked that question and then you answer it in 17 the negative, yes? 18 A. Err, yes -- 19 Q. Or at least bordering on the impossible, you 20 say, extremely unlikely? 21 A. Yes, and then I carry on discussing it in 22 paragraph 19 as well. 23 Q. You do indeed. If we go down to paragraph 21, 24 you make a general observation: 25 "... counter replacement, though it could 123 1 potentially result in transactions being lost if 2 the broken counter had been operational but 3 disconnected prior to replacement, would not be 4 a cause of losses occurring two or three weeks 5 later." 6 So you do acknowledge there that the 7 possibility of losing transactions is present 8 when there's a failure of a hard disk or a node. 9 A. There's a possibility, yes. 10 Q. That's something that you knew was not just 11 a possibility but sometimes happened it; is that 12 right? 13 A. Yes, it did happen but -- yes, it did happen but 14 only if you had other problems, comms problems, 15 as well as the counter replacement. 16 Q. Well, there's one example, isn't there, in -- 17 I can take you to the document to help us deal 18 with this. It's POL00000994. This appears to 19 have been a reasonably significant loss of data. 20 We can see there at the top "Recovery of 21 overwritten transactions", and we get the code 22 for the branch where this happened. 23 A. Yes. 24 Q. If we go down a little bit and we look under 25 "Extra detail", we can see a bit of 124 1 an explanation of what happened. 2 A. Yes. 3 Q. "Due to problems with the gateway, which has 4 been replaced twice, and communications ..." 5 So that confirms what you were saying: that 6 it's often when there's a communications issue 7 as well, yes? 8 A. And, in this case, it was communications both 9 between the gateway and the data centre and 10 between the gateway and the other counters, so 11 it was two different sets of communications. 12 Q. But what we do see though is that approximately 13 900 transactions -- this is the second 14 paragraph -- done on the gateway between 1300 on 15 the 9th and 11.46 on the 12th were effectively 16 no longer present; is that fair? 17 A. Yes, they were no longer in the branch message 18 store. 19 Q. Right, and: 20 "To get the messages back, we will [need to] 21 delete the counter message stores and let the 22 version from the correspondence server replicate 23 down." 24 Yes? 25 A. Yes. 125 1 Q. Then it goes on to say at the end: 2 "We will reinsert the transactions which 3 were completed in that time, and reset the stock 4 unit ..." 5 Yes? 6 A. We reinserted the messages written on 7 and 7 8 September, yes. 8 Q. So this is an example of messages being lost 9 following hardware failure? 10 A. Yes. 11 Q. All right, well, we can take that down. 12 Is it because of that sort of issue that you 13 thought to yourself, when making your statement, 14 "Well, I see that there was a hardware 15 replacement at Marine Drive, I'd better sort of 16 cover that off, if you like?" 17 A. It was knowing that Mr Castleton had asked 18 questions in this area in court in 2006, which 19 I felt I perhaps hadn't answered as clearly as 20 I should have done at the time, partly because 21 I wasn't expecting to be asked about it. So 22 I felt it was worth trying to spell out in my 23 statement here my view of the situation, based 24 on what I could remember from them and also 25 examining bits of evidence that have been made 126 1 available to me in the run-up to the -- this 2 appearance. 3 Q. Am I right in saying there was an overnight 4 process checking across the whole Post Office 5 estate for hardware failures, "bad blocks", as 6 I think it was termed? 7 A. I think there was something like that. I can't 8 remember details. 9 Q. Do you know when that was introduced or why? 10 A. No, I don't know. That wasn't something that 11 SSC were responsible for. 12 Q. No, all right. There was a sort of an issue, if 13 you like, around recovery after hardware 14 failures, and the need to make sure that any 15 transactions that had been affected by hardware 16 failures needed to be checked over, if you like, 17 and made sure that they were present? 18 A. Yes, I mean, recovery in the counter sense is -- 19 would really just be any transactions that were 20 on the stack at the point of failure that didn't 21 get settled/written to message store. 22 Q. If we look at how that's handled in the Horizon 23 user guide, we can see that in POL00071234. 24 Page 29 is where a longish section begins, 25 headed "Equipment failure checklist (dealing 127 1 with equipment failure)", and then we see at the 2 top of the page it says, "System failure". That 3 section 12 is reasonably substantial. Do you 4 recognise this? 5 A. I had that -- I had access to the Horizon System 6 User Guide. I cannot remember whether I ever 7 read this particular section. 8 Q. If we go down to page 40, "Identifying lost 9 transactions following a system failure", this 10 is the section, still, as you see, headed 11 "System failure", so we're still dealing with 12 system failure as the overarching subject, and 13 identifying lost transactions. 14 Then if we scroll down, we can see that 15 a very substantial flowchart is presented, 16 "Recovery procedures after system failure". We 17 can see this flowchart. If we keep scrolling 18 down, it does rather go on and on. If we just 19 carry on, you'll see what I mean. You see how 20 a subpostmaster was supposed to go about dealing 21 with matters of recovery after a hardware 22 failure. 23 We can effect take that down, I don't 24 propose to take you to the detail, but let's 25 suppose that, for whatever reason, 128 1 a subpostmaster was not able to work through 2 that flowchart perfectly, presumably that may 3 result in lost or missing transactions? 4 A. Without spending a lot of time going through 5 that flowchart, which I'm sorry but I haven't 6 ever examined in any detail, I don't think I'm 7 going to be able to answer that. Basically, if 8 there was a set of transactions on the stack and 9 the counter fails at that point, without 10 settling them, then they will be thrown away but 11 there are certain extra things that needed to be 12 done, particularly for if it -- there were any 13 bill payments in that set of transactions. 14 Q. Which might pick it up in reconciliation; is 15 that right? 16 A. That might be picked up in reconciliation. 17 Without seeing a particular example, it's hard 18 to know. Banking transactions would have been 19 sort of picked up if necessary through 20 reconciliation and there was something at the 21 end there about if what was on the stack was 22 a rem in or rem out, that might have an effect. 23 But I have to say I don't see the relevance of 24 that to the failure overnight on a Saturday 25 night at Marine Drive. 129 1 Q. Well, that was directing subpostmasters to 2 a process that they were supposed to complete if 3 there'd been a system failure -- I mean, 4 a hardware failure. So, presumably, that was 5 what any subpostmaster presented with a hardware 6 failure was supposed to go through? 7 A. But wasn't the first question, "Were you serving 8 at the time"? 9 Q. What relevance would that have? 10 A. Because if you weren't using the counter at the 11 point that it failed, then you -- if you'd 12 logged out successfully and the counter was not 13 in use, then there aren't going to be any 14 transactions on the stack that haven't been 15 settled, because you can't log out without 16 settling the stack first. 17 Q. Right. So it would only be affected, would it, 18 if there were settlements currently going on? 19 A. Yes, if a postmaster was serving, middle of the 20 day, power cut, everything dies, in that case -- 21 or for some reason the screen freezes, you can't 22 get any further -- that situation, where you 23 were serving, then you would need to consider 24 whether recovery is going to be necessary when 25 you're able to log back on, either to that same 130 1 counter or to a replacement one. But if it 2 fails in the middle of the night, after you've 3 finished your trading, you're not using it, 4 there can't be any incomplete transactions, 5 recovery is not an issue. 6 Q. Can I just ask a question of Mr Castleton? 7 A. Yes, certainly. (Pause) 8 Q. Where there's a screen freeze or a failure 9 mid-transaction, then that's something that the 10 subpostmaster would need to go through, yes? If 11 that didn't work -- and this is not necessarily 12 talking about the overnight failure that we were 13 talking about before -- but if that didn't work, 14 there's the potential for the subpostmaster to 15 fail to be able to get back their incomplete 16 transactions, yes? 17 A. Yes, if you've got a screen freeze in the middle 18 of the day while you're serving then, yes, 19 anything that was on the stack but not settled, 20 which should just be transactions for a single 21 customer, then they will not exist. 22 Q. So if they had failed to get through that rather 23 lengthy flowchart and not been able to 24 successfully bring the transactions back through 25 doing that, would that be deemed user error? 131 1 A. I'm not sure if anybody would ever seriously 2 have used that flowchart. What would happen 3 when somebody tried to log back -- when somebody 4 was able to log back on to the counter again, if 5 there were bill payments, if there had been 6 a bill payment, which had been made but not 7 settled, then the postmaster would actually have 8 been prompted to say whether it's -- the bill 9 had been paid or not, and that would have 10 taken -- that should have sorted that out. 11 Possibly banking transactions as well but 12 I can't properly remember. 13 If there were other things on the stack, for 14 example a pension -- I mean, the postmasters 15 I think were expected to know that -- but they 16 might not have done -- that if it failed before 17 settlement, they shouldn't have paid out the 18 pension money, or whatever. But even if they 19 didn't know that, if they had paid out the money 20 but it hadn't settled, then, at the end of the 21 day when they checked all their counterfoils 22 against the report -- the totals on the report, 23 that should have been apparent as a difference, 24 and they could then have put it through at 25 a later stage. 132 1 Q. But can I just understand this: if, for whatever 2 reason, the postmaster was struggling with this 3 and there remained a problem, would that be 4 ascribed as "user error"? 5 A. I'm not sure who would be doing that ascribing. 6 Q. All right. Well, let's look at the instance of 7 Mr Booth, who was the second temporary 8 subpostmaster at Marine Drive, who came in after 9 Mr Castleton was suspended. He got in touch 10 with the Post Office's lawyers to let them know 11 he'd experienced lost transactions when working 12 in the branch, as a result of a screen freeze. 13 Ultimately, it was presented in the case of POL 14 v Castleton as the system had worked as it 15 should and that any problem with disappearing 16 transactions was, in effect, a failure by 17 Mr Booth? 18 A. Yes, I don't think it should have been 19 categorised in that way. 20 Q. Well, can we just look at a sort of a snippet 21 from a document, which is POL00081826_022. If 22 we go to page 7, please. When the lawyers were 23 dealing with this and they were back and forth 24 with Fujitsu, specifically Brian Pinder, this 25 message at the top here -- so we can scroll up 133 1 just to sort of contextualise it -- on 2 2 November, this is an email from Mr Pinder to 3 Mr Dilley, with various others, the legal team, 4 in effect, copied in, and it says: 5 "Stephen. 6 "[I'm guessing] You might wish to note that 7 ..." 8 Then it appears that he's quoting from 9 something: 10 "Should the system be restarted (for any 11 reason -- including following a 'freeze'), there 12 will be evidence of this in the Audit trail 13 (which we have in fact been examining in this 14 case). Normally the only system restarts are as 15 part of the overnight 'ClearDesk' function that 16 occurs between 03.30 and 04.00 each day. Any 17 other restarts can be considered unusual and 18 could be searched for." 19 So if we just pick that apart, I suggest -- 20 and you can correct me if I'm wrong -- that the 21 overnight ClearDesk function is what we spoke 22 about earlier, that there was an estate-wide 23 check for any problems with disks. 24 A. No, that's not what ClearDesk was. 25 Q. No? 134 1 A. The Riposte service on every counter closed down 2 and restarted at some point between 3.30 and 3 4.00. 4 Q. Okay, so that was everywhere, was it? 5 A. That was all counters, yes. 6 Q. All right. But, apart from that, which happened 7 every single night everywhere, any other 8 restarts can be considered unusual and could be 9 searched for? 10 A. Yes. 11 Q. It appears from the section in bold that that 12 was being undertaken. Do you see what I mean? 13 A. Yes, I see what you mean. I'm just looking at 14 dates. Yes, I have no recollection of 15 specifically looking for those things but, in -- 16 so it could have happened, it may have happened. 17 I just can't remember. 18 Q. So you just don't know if that was you that was 19 trying to carry out that check for unusual 20 restarts? 21 A. Yeah, I don't remember, I could have done it. 22 You wouldn't see it so easily in the ARQ data, 23 but you would see it in the full message 24 extract. 25 Q. Given what you said to Mr Beer earlier about 135 1 working documents, and so forth, there's now no 2 way, really, of knowing if there were any 3 unusual reboots, is there? 4 A. The only other way you could do -- and I did try 5 looking a little bit the other day -- would be 6 to see if you've got log-ons with no matching 7 log-offs happening. That would be an indication 8 that the counter was restarted but that's the 9 only thing you could check for in what we have 10 left to us now. 11 Q. We do know that that applied in the case of the 12 overnight problem in early February, don't we? 13 A. No, that was a different case where there was no 14 log-off when there was a log-on of the same user 15 on a different counter. That was a different 16 thing. That wasn't a failure to log off. 17 Right. In that case, I could -- the log-off 18 was done successfully the night before the 19 counter failed and then, when we were looking at 20 transactions on 2 February, we could see that 21 Mr Castleton logged on and was using -- serving 22 on counter 1, and then the same user logged on 23 to counter 2, without logging off counter 1 but, 24 in that case, it doesn't appear that counter 1 25 became unusable, at that point. He just chose 136 1 to move from one system to the other. 2 Q. Well, just standing back and taking the view 3 overall, if we may. Whatever examination there 4 was for unusual restarts in Marine Drive we now 5 don't really have any conclusive evidence of it; 6 is that right? 7 A. There's -- I haven't been able to find any 8 evidence that it was happening but it is 9 a possibility that there were some and no, we 10 don't have anything that would be conclusive. 11 Q. All right. What that could have shown is 12 whether there were problems with screen freezes 13 and, therefore, potentially missing data 14 following screen freezes or it would have been 15 easier to find the possibility, if you'd found 16 unusual restarts; is that right? 17 A. Yes. But, as I said, it wouldn't necessarily 18 cause missing data, but it might -- 19 Q. But it might do. 20 A. Potentially, there might have been sessions that 21 didn't settle but that wouldn't necessarily 22 cause discrepancies. 23 Q. No, but it might do? 24 A. Unlikely to but, yeah, it would depend on the 25 individual circumstances. 137 1 Q. All right. Going to the questions that you were 2 asked by Mr Castleton in the trial in 2006, we 3 can have a look at those at LCAS0001300. If we 4 go to page 271 and down to the bottom of that 5 page. If we just scroll down, the questions had 6 already started but the bit that I was going to 7 focus on is this: 8 "Question: But it does not show the 9 reconnection before disconnection again, does 10 it? 11 "Answer: Whether it was because the 12 engineer was briefly plugging things in and then 13 unplugging them again --" 14 If we scroll down a little: 15 "Question: It does actually occur in other 16 areas. 17 "Answer: -- I really cannot explain that, 18 not from this information. If I looked at it in 19 conjunction with other things --" 20 That's you then being a little taken by 21 surprise, is it? 22 A. Yes. 23 Q. So you're effectively saying there "I'm not 24 sure, the engineer may have been doing X, Y and 25 Z but I can't be sure, based on what I've got in 138 1 front of me"? 2 A. Yes. 3 Q. There is a little similarity here with some of 4 the other evidence you've given, even just the 5 evidence you've given just now, which is to say, 6 "I can't be sure, based on all the material that 7 I have in front of me now. There is potentially 8 other material", but also you tend to offer 9 an explanation, such as the explanation about 10 the engineer unplugging things or the 11 explanations that you've offered today about the 12 transactions being ones which might have been 13 picked up in reconciliations and so forth. 14 Your explanations or your hypotheses that 15 you put forward tend to explain it in terms of 16 supporting the system, in terms of making sure 17 that whatever problems there may have been, 18 there probably weren't, because of X, Y and Z; 19 do you agree? 20 A. Yes, I think that's a fair comment, but it's 21 based around 20 years, well, 15 years' knowledge 22 of how the system worked and a general feeling. 23 But, yes, I like to explain things and, you 24 know, in the past, I would have been able to 25 then back that up by looking at stuff that would 139 1 support it and, yes, I cannot do that any more 2 because we haven't got the evidence. 3 Q. All right. Well, the fact is, when branches had 4 hardware failure, which lost data, people in SSC 5 could insert it, couldn't they? 6 A. As a last resort, yes. That would be done in 7 order for the branch to balance correctly, 8 including transactions which they had carried 9 out but had failed to get to the places where 10 they should have been. 11 Q. The claim, as you put in your witness 12 statement -- and we don't necessarily need to go 13 there -- is that, when that happened, the SSC 14 were intending to put the branch accounts into 15 the -- or rather intending to put the system 16 into the state it would have been in in the 17 first place, if it hadn't had the failure. 18 A. Yes. 19 Q. What if the system problem was imperfectly 20 understood or imperfectly corrected? Sometimes, 21 the subpostmaster had no idea what was going on, 22 did they? 23 A. I can certainly think of one occasion when that 24 is true, yes. 25 Q. So -- 140 1 A. But it -- yeah. 2 Q. -- in the circumstances where -- and, let's face 3 it, people in SSC are human beings and they 4 don't necessarily get it right -- in the 5 circumstances where someone in SSC has got it 6 wrong and the SPM, the subpostmaster, has no 7 idea of what has been happening, there would be 8 no way of identifying where the problems began 9 or ended, would there? 10 A. Do you mean no way of the postmaster identifying 11 that or for anybody else identifying that? 12 Q. Well, certainly for the subpostmaster and 13 I think you would have to agree with that; is 14 that right? 15 A. I think I would agree with that, yes. 16 Q. The only way anybody would be able to get to the 17 bottom of it at the other end is if they had the 18 raw data and if the person in SSC had left the 19 right sort of notes behind for people to know 20 what they'd done; is that right? 21 A. Yes, I think that's right. I would say that 22 these changes were extremely unusual. It was 23 certainly not something that was being done with 24 any regularity at all but probably a few times 25 a year, over all 18,000 branches, it was 141 1 necessary to do -- it was felt necessary to do 2 some sort of correction. 3 Q. Is it because of the rarity of it that you do 4 not, effectively, countenance the possibility 5 that the pension transactions that you were 6 looking at with Mr Beer were inserted by SSC in 7 an attempt to rectify data that may have been 8 lost? 9 A. I totally discount that explanation because you 10 can see that they took place on counter 1, you 11 can see the session ID number. You can see that 12 Mr Castleton, or whoever was using LCA001, was 13 serving on counter 1. He did a whole set of 14 pension transactions in one session, possibly 15 catching up for things that had been paid out 16 manually during the morning because they'd been 17 reduced to working on a single counter. 18 That session was then suspended, node 2 came 19 back up Mr Castleton logged on to node 2, which 20 transferred these -- this suspended session 21 across from counter 1 and then you can see that 22 that session was eventually unsuspended and was 23 settled on counter 2. And that is quite clear 24 from the transaction numbers and the session 25 numbers, and anybody inserting those afterwards 142 1 would not have been able to use those numbers 2 because they were -- the sequence numbers were 3 being used by the person on the counter, logged 4 in on the counter at that time. 5 Q. All right. Disclosure. One short topic before 6 I have a final topic after that. 7 So on disclosure -- this won't take long -- 8 was the extract from the message store that you 9 produced to do your report on week 42, was that 10 the full message store for the Marine Drive 11 branch? 12 A. It was the full set of messages in that 13 particular week. 14 Q. Oh, just for that week, all right. 15 A. It was just for that week was -- 16 Q. So it wasn't for the whole period; it was just 17 week 42? 18 A. That was what I looked at when producing the 19 week 42 cash account or when attempting to. 20 But, subsequently, I am just about certain that 21 we did have all the messages for January and 22 February, which I did then look through. But 23 I have not got any -- 24 Q. Any proof of that? 25 A. I haven't seen anything to prove that. 143 1 Q. No. 2 A. But I know I spent a lot of time sitting there 3 and looking at the data over that whole period. 4 Q. The extract that we know did happen, because 5 otherwise you couldn't have written that report, 6 the week 42 extract, just thinking about the 7 disclosure explanation that you had from 8 Mr Dilley at the end of the case that you quoted 9 in your afterthoughts, really that should have 10 been disclosed, shouldn't it? 11 A. Yes. 12 Q. Was that ever passed to Mr Dilley with the 13 report; do you know? 14 A. I don't know because I was not responsible for 15 disclosure. There was something in that comment 16 that I quoted from Mr Dilley saying that he had 17 had sight of the message store and I can't 18 remember if it's in there or another document 19 from Mr Dilley saying that he'd offered 20 Mr Castleton sight of the message store but -- 21 Q. Yes, certainly he'd offered him sight of the 22 message store. 23 A. Yes, but because it was so big and difficult -- 24 and I don't know if that was just week 42 or if 25 that was for the entire period. 144 1 Q. But you can't be sure whether you gave the 2 extract of week 42 to Mr Dilley, you just don't 3 know? 4 A. Me, personally, no, I didn't. 5 Q. No. You didn't -- 6 A. No. 7 Q. -- I see. 8 SIR WYN WILLIAMS: Can I ask you, were you ever 9 asked a direct question by Mr Dilley, or anybody 10 I imagine, but let's say by anyone: during the 11 course of your investigations, Mrs Chambers, 12 have you created any documents which you 13 consider might be relevant to the issues we are 14 considering in this trial? 15 A. I don't recall anybody ever asking me that 16 question, no. 17 SIR WYN WILLIAMS: All right. Thank you. 18 MS PAGE: Then just your explanation about KELs and 19 being told not to mention KELs and, indeed, to 20 positively refer to other documents, rather than 21 KELs. Again, that's something which, looking 22 back, it would have met that disclosure 23 obligation that Mr Dilley set out for you and 24 you set out in your afterthoughts, wouldn't it? 25 A. Yes. 145 1 Q. All right. Now, why did accept that you 2 shouldn't refer to KELs? 3 A. I assumed that -- that's what they said and so 4 that's what I did. Perhaps I should have 5 questioned it but I was in a very unfamiliar 6 situation for me, and -- 7 Q. Why didn't you mention it in your afterthoughts 8 document? Because, at that stage, obviously, 9 you had been told by Mr Dilley what you'd been 10 told about your disclosure obligations and, as 11 we've just agreed, KELs would have been 12 disclosable under that test. So why not mention 13 the problem with you being told that you 14 shouldn't refer to KELs in your afterthoughts? 15 A. I thought I did. Did I mention PEAKs in there? 16 Q. You certainly talked about including things like 17 the event logs on the back of PEAKs and making 18 sure that people had PEAKs, but not KELs. 19 A. Yes, that was an oversight on my part when 20 I wrote the afterthoughts document, then. 21 I mean, I perhaps thought that PEAKs, you 22 know -- it wasn't just the KELs that weren't 23 being disclosed; it was PEAKs, which I think, to 24 me, I thought was important as well but, yes, 25 maybe KELs equally so, but I didn't include it 146 1 in that document, but for no reason. 2 Q. Did it not strike you even then as perhaps 3 slightly suspicious that nobody wanted you to 4 mention known error logs with that title being 5 what it was? 6 A. I don't think I thought of it as suspicious. 7 I thought it seemed strange but, as I said, 8 I was in a very unfamiliar situation. 9 Q. All right. Well, finally, a few questions on 10 what I think will be just a short final topic. 11 Did you believe that Horizon was fundamentally 12 robust in 2004 -- 13 A. Fundamentally, yes. 14 Q. -- and still in 2006? 15 A. Yes. 16 Q. Confirmation bias happens when we filter all the 17 evidence we receive to only take account of 18 those parts which support what we already 19 believe to be true. Are you familiar with that 20 concept? 21 A. I'm familiar with the concept, yes. 22 Q. When you gave evidence in 2006, do you think you 23 were affected by confirmation bias, your belief 24 in the fundamental robustness of Horizon? 25 A. No, because I knew I had looked in great deal at 147 1 the available evidence, and I was looking for 2 any sign of a system problem and the numbers 3 added up. It wasn't that the discrepancies were 4 being calculated wrongly. 5 Q. But you believed the system to be fundamentally 6 robust, didn't you? 7 A. No, not necessarily. No, I mean, if there had 8 been any sign that the numbers were not adding 9 up, then I would have been onto it and I would 10 have said so. But the discrepancies were 11 calculated correctly, based on the transactions 12 that had been recorded. There were no obviously 13 incorrect transactions in what I could see of 14 the system. 15 All I could conclude was that what was on 16 the system was not an accurate reflection of 17 what had taken place at the branch and there was 18 nothing that indicated that the system had 19 malfunctioned in any way that would give rise to 20 transactions either not being on the system or 21 being there but there incorrectly. But, 22 fundamentally, that -- finally, that could only 23 be ascertained by comparing what was on the 24 system with what had actually taken place at the 25 branch and once the day/the week has passed, 148 1 that becomes almost impossible to verify. 2 Q. Are you still, at some level, trying to cover 3 off or explain away those anomalies or issues 4 which we can see? 5 A. No, where there have been anomalies, for example 6 the missing 92p in the ARQ data, I've found 7 that, I've investigated it, I've checked what 8 the consequences of that are and I have reported 9 it. Whether that report didn't get through to 10 Mr Castleton and his team, I don't know, because 11 I was not reporting direct to him, but I had no 12 intention of covering that up. 13 Q. Do you now believe that the case against 14 Mr Castleton resulted in a miscarriage of 15 justice? 16 A. I think Mr Castleton should have been given much 17 more help in trying to work out why the money in 18 his branch did not match what was on the system, 19 and I wish, really wish, I could find a problem. 20 I've still been looking to see if there's just 21 anything that looks as if it could be something 22 wrong with the system that might have caused it 23 but I found nothing in 2004. 24 And, okay, you can say I didn't look hard 25 enough then. I did what I -- I didn't know that 149 1 this would still be being talked about. 2 Q. No. 3 A. I did a reasonable job in 2004, which I then 4 gave evidence on. I rechecked, looking for 5 errors, not looking for explanations of why 6 there weren't errors, in 2006. I continued to 7 wonder if there was something I had missed. If 8 it was, what could it be? I've never found any 9 error in Horizon -- Legacy Horizon that could 10 have caused Mr Castleton's losses and, so, 11 yes -- 12 Q. You now, do you, say that if there was an error 13 in Horizon, it's one that you simply didn't 14 find? 15 A. It's one that had -- that left no evidence and 16 that nobody has ever found, if there was such 17 an error, if there could even be such an error. 18 MS PAGE: Thank you. Those are my questions. 19 SIR WYN WILLIAMS: Thank you, Ms Page. 20 Is that it, Mr Beer? 21 MR BEER: Yes, it is, sir, unless you have any 22 questions? 23 SIR WYN WILLIAMS: No, thank you. 24 It's obvious that you have provided two very 25 detailed witness statements, Mrs Chambers, and 150 1 you've given evidence before me over very many 2 hours. For all those things, I thank you very 3 much. 4 MR BEER: Thank you very much, sir. We reconvene at 5 10.00 tomorrow morning. 6 SIR WYN WILLIAMS: Yes. 7 MR BEER: Thank you. 8 (2.56 pm) 9 (The hearing adjourned until 10.00 am 10 the following day) 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 151 I N D E X ANNE OLIVIA CHAMBERS (continued) ..............1 Questioned by MR BEER (continued) .............1 Questioned by MS PAGE .......................119 152