SUBS0000027
Post Office Horizon IT Inquiry
On behalf of Core Participants represented by Hodge Jones & Allen: Teju Adedayo,
Nichola Arch, Lee Castleton, Tracy Felstead, Parmod Kalia, Mhairi McDougall,
Seema Misra, Vijay Parekh, Vipin Patel, Colin Savage, Sathyan Shiju and Janet Skinner
Closing Statement Phase 4 - Action against Sub-postmasters and others: policy making,
audits and investigations, civil and criminal proceedings, knowledge of and
responsibility for failures in investigation and disclosure
Introduction sates
The relevant law: Disclosure; Conspiracy to pervert the course of public justice...
The POL v Castleton conspiracies...
The POL group conspiracy...
Keith Baines’s knowledge .
Corporate dissemination of Mr Baines’s knowledge
The legal ‘sleight of hand’ .
Misconduct by POL’s lawyers...
The Fujitsu group conspiracy.
Knowledge within the Customer Services Directorate
Fujitsu’s failure to understand the role of litigation support
Using Anne Chambers to avoid exposing Fujitsu's failures ....
The response to Chambers’ Afterthoughts ...
Anne Chambers’ perjury ...
Stephen Dilley’s failures of disclosure and the drafting of Statements...
The criminal prosecutions conspiracy .....
The ad hoc contractual arrangements for POL prosecutions ...
Jan Holmes’ role in Cleveleys and the prosecution of Tracy Felstead..
Symbiosis: Fujitsu (Holmes) and POL...
Designing the litigation support service ....
Remote access...
The 2008 ARQ problems...
POL Security was drenched in information about Horizon errors...
Doctoring the Noel Thomas expert statement ....
SUBS0000027
Over-charging and extracting confessions...
Systematising perverting the course of justice.
PRECIS OF DUNCAN ATKINSON’S KC EVIDENCE
The R v Seema Misra conspiracy
Stonewalling the defence requests for evidence and disclosure...
Abuse of expert evidence ....
Wilson’s intervention in the Ismay report ....
The Receipts & Payments Mismatch bug ..
Absence of Cross-Disclosure: the Hosi case . 38
The “bandwagon” email and senior reaction to it
Preview of the cover-up conspiracy...
Conclusion
Introduction
1. The Post Office’s heritage is now a thing of shame. Phase 4 has uncovered a deplorable
culture of callous abuse. SPMs with shortfalls were bled under a draconian contract,
where the one-sided contractual position for losses was made more intolerable by POL’s
ruthless pursuit of cash. This oppression mutated into an unprincipled default
prosecution policy for gain. Investigators sometimes did not even believe the guilt of
those they went after. Vital evidential and public interest concerns were ignored in POL’s
all-consuming appetite to prosecute. As Mr Duncan Atkinson KC stated:
“rather than using civil recovery under the contract, they were using the criminal process and
the levers of the criminal process, such as confiscation, such as it being a condition of the
acceptance of a plea to get the money back, when they hadn't actualy proved that the money
had gone in the first place.”!
2. Underlying all of this, was something profoundly more sinister: a symbiotic covenant of
secrecy between POL and Fujitsu. The Inquiry has seen a Fortress Horizon mentality
upon which reasonable requests for disclosure were dashed, and from which no
disclosure willingly emanated. The Inquiry has witnessed the wholesale disapplication
of the CPIA, and The Attorney General’s Guidelines on Disclosure, the manipulation and
re-writing of expert evidence, the destruction of documents, and the suppression of the
truth that could have averted each wrongful conviction it has been tasked to investigate.
Civil and Criminal justice was pitilessly subverted to serve POL’s commercial interests.
Collectively, a nadir has been reached and yet there is worse to come, for the culture (set
from the top) that fostered such abuses, allowed no supervisory oversight to ensure best
practice and so restrain arbitrary misuse of power or wrongdoing.
3. The reputation of British justice has, quite literally, been shredded. To restore confidence,
and to uphold standards across this dismal scene, the Inquiry might consider it
appropriate to indicate where it considers further investigation is merited where offences
may have been committed, or where serious professional misconduct may have occurred.
+ Evidence of Duncan Atkinson KC, 18 December 2023 p.44/ Line 25 to p.46/line 5
2
SUBS0000027
This is not to impute liability, whether civil or criminal, but to recognise the obvious,
inevitable and inescapable consequences of the evidence it has heard. The Inquiry has
such powers, and the Core Participants we represent have confidence that such
indications, if not formal recommendations, will follow in due course.
The relevant law: Disclosure; Conspiracy to pervert the course of public justice
4, The law in relation to Disclosure is taken from the reports and the evidence of Duncan
Atkinson KC, summarised below at §§ 154-159.
5. We rely on the law of conspiracy to pervert the course of justice, as set out in our Phase 3
Closing Statement, and on imputed knowledge, as set out in our Phase 2 Closing
Statement?
The POL v Castleton conspiracies
6. Mr Castleton’s life, reputation, livelihood, and the wellbeing of his family was sacrificed
to protect the infallibility of Horizon. By the time of his trial it was known from the
Cleveley’s case, which preceded it, that there were problems with system errors. These
were suppressed and withheld by an NDA and were never disclosed. Mr Castleton was
destined to become a victim in POL’s ruthless determination to acquire a precedent. We,
therefore, submit that there is evidence to support the following allegations:
a. Keith Baines, Mandy Talbot, Tony Utting, Graham Ward, Rod Ismay and others (the
POL group) conspired together, and with others to pervert the course of justice by
pursuing a debt claim on behalf of POL against Lee Castleton’
b. Naomi Elliott and Brian Pinder (the Fujitsu group) conspired together, and with
others, to pervert the course of justice by distorting and concealing evidence relevant
to the claim being brought by POL against Lee Castleton
c. Anne Chambers committed perjury when testifying in POL v Castleton, by agreeing
not to mention the Known Error Log, and obscuring the potential for there to have
been errors in the Marine Drive cash accounts.
The POL group conspiracy
7. Given the evidence from the Bates litigation and Phase 1 of the Inquiry that Horizon could
and did produce unreliable cash accounts, POL’s reliance upon the Marine Drive cash
accounts during the Castleton litigation, knowing there was expert evidence as to its
unreliability’, was a course of conduct which had a tendency to pervert the course of
justice. It allowed POL to obtain judgment for a debt, and the consequent costs order,
without producing reliable evidence that the debt was owed.
8. There is evidence that this course of conduct was also maliciously intended to pervert the
course of justice, because the POL group either knew that the Marine Drive cash accounts
may not have been reliable, or they closed their eyes to that knowledge. The motive for
the course of conduct may have been to protect POL, or their own jobs, and the natural
consequence of the course of conduct may not have been desired as such, but that does
not mean it was not intended. Mr Castleton’s destruction may have been seen as
unfortunate collateral damage in order to protect Horizon at all costs.
? SUBS0000018 §2; SUBSO0000022 §§9-13
3 The rationalisation to resist a counterclaim by Mr Castleton is a distinction without a difference, as his claim
was predicated on the injustice inflicted upon him by Horizon. The importance of his counterclaim dwindled
during the run up to trial.
# Jason Coyne’s joint expert report in Cleveley’s
SUBS0000027
Keith Baines’s knowledge
9. The late Keith Baines, Contract Manager, was intimately involved with Acceptance
Incident 376 (unreliable cash accounts). Documents show that he managed the
negotiations around the consequent suspension of the rollout, and the Second
Supplemental Agreement. Under Schedule 4 Part B (i) of that Agreement, ongoing
problems with the cash accounts were settled by instituting a monitoring period of six
weeks in late 1999. If, during that period, cash accounts with discrepancies did not exceed
0.6%, AI 376 would be resolved.$ That meant that a small number of discrepancies would
not be a bar to Horizon rolling out, so some unfortunate branches in the network would
be highly likely to experience discrepancies in their accounts.
10. Then in mid-2001, Mr Baines managed the contractual resolution of a significant problem
with ARQ data. Jan Holmes, of Fujitsu, notified the Post Office that from 8 to 14 August
2000 there was a break in the available ARQ data, due to corruption of both tapes which
should have been storing it? This put Fujitsu into a position of straightforward
contractual non-compliance until 15 February 2002, due to the provision which required
retention of audit data for a period of 18 months. However, on 7 August 2001, on behalf
of Fujitsu, Colin Lenton-Smith informed Keith Baines that “measures to remove
altogether the risk of future tape corruption can be achieved only by a complete re-design
of the current solution”.§ That led in due course, after a little back-and forth, to Keith
Baines settling for a “goodwill” payment from Fujitsu of £150,000? So again, Mr Baines
accepted a resolution which left open the possibility of future problems, this time
compromising the accuracy of ARQ data, which was supposed to demonstrate the
reliability of Horizon cash accounts.
11. These failures by Mr Baines may have left him feeling exposed when on 20 January 2004
Jason Coyne reported on the Cleveleys case, and found that:
From the 31st of October... there seems to be a numberof [helpdesk call] logs which talk of “large
discrepancies” in stock figures, trial balances with “all sorts of figures showing minus figures”.
From a computer system installation perspective it is myopinion that the technology installed at
the Cleveleys sub-post office was clearly defective in elements of its hardware, software or
interfaces. The majority of the errors as noted in the fault logs could not be attributed to being of
Mrs Wolstenholme's making or operation of the system. '°
12. Mrs Wolstenholme difficulties in November 2000 came very soon after Mr Baines had
agreed that AI 376 could be resolved, despite the potential for some cash accounts to have
discrepancies. Yet on 14 October 2003, he signed a witness statement, saying this:
Any faults that occurred in the Horizon computer system were eliminated once they were
identified. Whilst it is possible for mistakes to occur, this is usually through incorrect inputting
to the computer system in the office affected by the mistake.All sub-postmasters were fully
trained in the use of the Horizon equipment. The system was fully tested before it was used by
the Post Office and it is fit for its purpose.The system itself does not create losses as is claimed
by Mrs Wolstenholme.
> POLO0028548, POLO0093313, POLO0028509
© FUjo0118149
7FUJ00171959
® FUJO0176280 p2
° FUJO0176298; FUJO0176299; FUJO0176300; FUJO0176301; FUJO0176303
20 WITNO9020115,
41 PQLO0118250, §5.
13.
14.
15.
16.
17.
SUBS0000027
Mr Baines knew that every sentence of this paragraph 5 was either untrue or misleading
by omission. On 26 July 2004, Counsel in the Cleveleys case made an explicit note that
POL was concerned to ensure that Mr Coyne’s Report did not receive publicity/? no
doubt because it gave the lie to the false statement Mr Baines had made the previous year.
On 3 August 2004, a note of a conference call with counsel indicates that Mr Baines was,
to some extent, jockeyed into providing a second witness statement." The note also makes
clear that Mr Baines was to draft the statement himself, and the fact that he faxed it to the
solicitor Susanne Helliwell, rather than the other way around, suggests that is exactly
what happened. In it he admits to “significant” problems at acceptance, and describes
some of them, but he makes no mention of AI376, which was obviously the most relevant
and important, nor does he mention the Second Supplemental Agreement permitting up
to 0.6% of cash accounts to have discrepancies."
He signed his second witness statement on 11 August, but importantly, on 4 August he
emailed it to Mandy Talbot, saying this:
This is general, rather than specific to Cleveleys, and in effect is the detail behind some of the
assertions in paragraph 5 of my earlier witness statement (signed on 14/10/2003)!5
At this point, Ms Talbot was holding the in-house brief in the Cleveleys litigation, but
also had carriage of the Castleton litigation, having taken it over on 7 June 2004.
Meanwhile, Graham Ward, a key person within Security, had informed Keith Baines of
another problem with ARQ data. On 1 June 2004, Ward forwarded an email from Bill
Mitchell at Fujitsu, which said that incomplete data had been sent to POL in a number of
cases which had gone to prosecution. On 15 June, Mr Ward forwarded his own email to
Mr Baines with an update, saying that a further fourbranches were impacted, and making
it clear that in these four and the previous two, the ARQ data had been used to underpin
prosecutions.
Corporate dissemination of Mr Baines’s knowledge
18.
19.
20.
Obviously, Mr Baines was not inclined to inform others in writing when he became aware
that AI376 and ARQ problems could be casting a shadow over legal proceedings in 2004.
However, there is evidence from which it can be inferred that others became aware.
An interesting email from Rod Ismay to Mandy Talbot (amongst others) dated 26 July
2004, shows that the POL hierarchy was fully engaged with the Cleveleys litigation and
settlement considerations by this stage. Ms Talbot asked Mr Ismay whether she should
now be taking her instructions from Peter Corbett, the Finance Director, rather than from
“Chesterfield”, to which he replied that he was escalating the case to Dave Miller, the
Managing Director. At the same time, he asked Tony Marsh to indicate who was leading
on the case within Security, and cautioned all addressees as follows:
please do not circulate this any further than is necessary to support Dave[Miller] and Group
Legal with this case. *
This ties in with the way POL expressed the risks associated withthe Cleveleys litigation
in its IT risk register:
+2 PQLO0118229 p4, and see also FUJOO121637, in which Mandy Talbot is quoted as saying she wants
to keep the Coyne report out of the public domain.
43 WITN04600310
14 POL00118224 — in citing this we observe that 0.6% seemed to be an underestimate in practice
45 POLO0118233
46 PQLO0142503
SUBS0000027
Damage to reputation of Post Office and potential future financial losses if PO loses court case
relating to reliability of Horizon accounting data at Cleveleys Branch Office!”
21. Already POL was more concerned with reputational fallout if word should get out that
Horizon accounting data was unreliable, rather than recognising that Cleveleys was a red
flag, requiring them to investigate whether Horizon accounting data was in fact reliable.
22. Ignoring that red flag brought its own problems. Over the course of 2005 a number of
other cases, including Castleton, came to the attention of the POL group, and, no doubt,
the hierarchy above them. In December 2005 Keith Baines, Mandy Talbot and Graham
Ward met with others, including Marie Cockett, Rod Ismay’s right hand woman in
Chesterfield. The issue under discussion was “Horizon integrity”, and it was noted that
There have been several recent cases where subpostmasters have cited errors in the Horizon
system as explanations for discrepanciesin their accounts...
Lawyers acting on behalf of a subpostmaster currently in dispute with Post Office have written
stating that they are contemplating a jointaction on behalf of a number of current and former
subpostmasters. This would challenge the accounting integrity of the Horizon system and Post
Office's right to make transaction corrections and recover resulting debts based on Horizon data.
In one past case (Cleveleys branch), Post Office settled out of courtfollowing an adverse report
on Horizon's potential to cause errors from an expert appointed by the court. Fujitsu advised that
the report was not well founded, but Post Office and Fujits u were not able to persuade the expert
to change it.'8
23. The meeting’s purpose was set out, and sensible actions were mooted, primarily co-
ordinating the POL and Fujitsu response to cases which raised the issue of Horizon’s
integrity, and appointing an independent expert to examine and report on Horizon data.””
The draft note of the meeting had assigned important actions to Keith Baines, in particular
“to discuss the need for and ToR of an external expert with Fujitsu”, and “to brief Dave
[X] Smith on the meeting’s recommendations” .”°
24. However, when Graham Ward’s boss, Tony Utting, was asked about the
recommendations set out in the meeting he said that he could not remember any of them
being put in place.”! He said that Rod Ismay would have approached Peter Corbett, the
Finance Director on “where to go with it”.
25. Itis telling that apparently neither Keith Baines nor Rod Ismay took the actions that were
expected of them to follow up on this meeting. During 2006 Rod Ismay encountered
further Horizon failings. We refer to §§36 to 39 of our Phase 3 Closing Statement for a
summary of the evidence he and Andrew Winn gave about the data migration problems
experienced within P&BA as a result of the Impact programme, and the “feeds from
branches” which were “falling into the wrong accounts” 5
26. Meanwhile, although Ms Talbot was not asked about the documents relating to December
2005 meeting, she was asked about one of her emails sent shortly before it which may
have prompted it. The email identified a putative class action from SPMs alleging that
Horizon caused discrepancies in their accounts, and she made suggestions which also
included coordinating the response to cases which raised the issue of Horizon’sintegrity
+7 PQLO0158493
+8 POLO0142539 p1
+9 POLO0142539 p2
2° POLOO119895 pp 4-5
21 https://www.postofficehorizoninquiry.org.uk/hearings/phase-4-17-november-2023 pp 83-95
22 Ibid pp 95-96
?3 SUBS0000022
SUBS0000027
and appointing an independent expert to examine and report on Horizon data.™ She too
said that none of her suggestions came to fruition:
Q. As at the end of 2005, going into 2006, 2007, was that independent report commissioned?
A. No.
Q. Were you concerned that it wasn't commissioned?
A. Of the, I think, five or so recommendations that I made, I was less concerned about the
commission of an independent report than I was about the creation of a team or even an
individual who would gather sufficient knowledge of the system so as to be able give proper
instructions on each case to external solicitors, so that we didn’t end up in another situation
where proceedings were issued without all the relevant material being present.
Q. Was that coordinating role established as at 2005/2006 and onwards?
A. ... ultimately, no such position was ever created?
27. Again, it is telling that Ms Talbot did not follow up on her own recommendations.
28. Ultimately, during the course of 2006, the POL group, or those above them, must have
29.
Th
0
30.
31.
32.
taken a decision not to instruct an independent expert, and not to coordinate information
about challenges to the integrity of Horizon accounts. This, we would submit, indicates
that Mr Baines warned others of the potential pitfalls in taking these steps. Rob Wilson
was less cautious when he wrote his March 2010 email,?° in which he once again deflected
colleagues from seeking a report from an independent expert, but even he did notreduce
to writing what must have been the real concern: an independent expert would find out
that Horizon was riddled with flaws, which would not reflect well on those who had been
supressing concerns for years, and carrying on prosecuting regardless.
This collective refusal to undertake the obvious and proper step of obtaining an
independent analysis of the Horizon system is the evidence which shows that the POL
group intended to deflect the Castleton litigation from finding out the truth about
Horizon; and therefore intended to pervert the course of justice.
legal ‘sleight of hand’
Richard Morgan KC’s reliance upon the case of Shaw v Picton” was the perfect answer to
POL’s problems and made it easier to pursue the conspiracy against Mr Castleton”. (We
refer to our Note re POL v Castleton, submitted on 25 September 2023, for an analysis of
Mr Morgan KC’s continued misunderstanding of the law applicable to “settled” accounts,
as determined by Fraser J in the Bates litigation.)
In legal terms, POL’s claim against Castleton had nothing to do with the reliability of the
Marine Drive cash accounts, or the reliability of Horizon accounting data more generally.
It was not a “test case” in the legal sense. The reliability of the cash accounts was solely
an evidential matter, just as all evidence is tested for reliability in any trial.
However, Ms Talbot recognised at an early stage that from POL’s point of view the only
significant issue in the case was Mr Castleton’s defence, which put the reliability of the
Horizon cash accounts in issue. In the email referred to above, dated 24 November 2005,
24 POL00107426 pS
25 https;,
www.postofficehorizoninquiry.org. i ? UOglijpp125-129
© POLOO106867
2748. &C.715
28 There is no suggestion that Mr Morgan KC wasa party, but his strategy concedes that Horizon was not error
free. This device wrought great injustice: Mr Castleton was required to sign off accounts he disputed and yet
the Court held him to have been voluntarily bound by his signature.Others, so compromised, suffered the
same fate, following Morgan’s disingenuous advice that they should be made to sign their (disputed) accounts
7
SUBS0000027
Ms Talbot said at the end of it: “If the challenge is not met the ability of POL to rely on
Horizon for data will be compromised and the future prosperity of the network
compromised.”?%
33. She also tied this challenge to what she saw as the need to be seen to be taking tough
action against Mr Castleton. In February 2006 Talbot apparently said to POL’s external
solicitor, Stephen Dilley, that ‘P.O must not show any weakness and even if this case
will cost a lot, there are broader issues at stake other than just Castleton's claim: if the P.O
are seen to compromise on Castleton, then "the whole system will come crashing down"
i.e it will egg on other sub postmasters to issue speculative claims.’°
34. However, in March 2006, she was still approaching this problem honestly. Her email to
Tony Utting, Marie Cockett, Graham Ward, David X Smith, Keith Baines and others on 1
March, began with a presumption that they would be following up the actions from the
December 2005 meeting, and she sought news on when someone would be appointed to
analyse Fujitsu data. She went on to say this:
I should be obliged for your comments upon what we believe that Fujitsu should beable to
provide by way of evidence and what they are obliged to provide under the contract. would
have thought that as a very minimum they should be able to say that they have run a check on
the whole network between 1/12/04 and 31/3/05 and can confirm hat either there were no
problems affecting the wholesystem, detail the ones which did occur, comment upon which
areas they affected and whether they would belikely to cause the problems complained of by
Castleton... I don't see any reason why Fujitsu coul dn't supply this information ... If Fujitsu
concur this can be built into the new process around investigation of these issues.
I would have thought that Fujitsu should be able to check the system with particular reference
to Marine Drive between the dates above and possibly afterwards to confirm whether or not they
have found any evidence of the problems complained of by Castleton... need to know whether
there's any justification for this allegation?"
35. This email met with resounding silence from the rest of the POL group. To the
extent she received replies she forwarded them to Stephen Dilley on 29 March with
the comment that the response to her email had been “limited in the extreme” *? As
time went by, Ms Talbot made no attempt to follow up on the reasonable (indeed
essential) requirements that she had suggested should be put to Fujitsu, reflecting an
earlier consensus.
36. On 16 August, Richard Morgan told the Bond Pearce lawyers that POL should be told
that it was “madness” to pursue the claim, given the costs that would be involved, and
that it could be settled with “drop hands” and a “confidentiality clause” ** Bond Pearce
were considerably less straightforward when passing that advice on. They merely sought
confirmation that their instructions were to pursue the claim, despite the knowledge that
the costs would “significantly exceed what is at stake”, and noted that therefore the
purpose was not to make a “net financial recovery”, but rather “to defend the Horizon
system” and “take a firm line” to “deter others from raising similar allegations” “4 They
did not raise the possibility of a confidentiality clause, no doubt because they knew that
POL were by now too invested in the idea of pursuing the claim to deter others.
22 POLO0107426 pS
3° POLO0070910 p1
31 POLOOO71202 pp9-12
2 POLO0071202 pé
33 POLO0072741 pé
34 POLO0082537 p2
37.
38.
39.
40.
41.
SUBS0000027
In the same email, and again at the behest of Mr Morgan, Bond Pearce sought
confirmation that POL/RMG did not know of any issues with the Fujitsu system or
Horizon integrity generally.2> Ms Talbot was asked why this had not prompted her to
pass on the Coyne Report from the Cleveleys case, and she said:
I was of the opinion that the preliminary view by Mr Coyne was created ina unique set of
circumstances, given that the original datawas no longer available. I didn’t consider it to bea full
report because the offer from Fujitsu for him to comeand visit their sites and look all over the
data was never communicated to him. So I didn’t consider that it was a full and comprehensive
report.
That answer is, we suggest, so plainly disingenuous, so plainly diversionary, that it can
be fairly inferred that by August 2006 Talbot was no longer approaching the Castleton
litigation honestly. She must have either drawn her own conclusions from the silence and
inaction following the December 2005 meeting, or someone had spoken to her, to make it
clear what was expected of her.
The latter possibility gains some support from the fact that whenever Ms Talbot was
asked about who within the Post Office gave her instructions in the Castleton litigation,
she affected complete amnesia. This may have appeared slightly more genuine if it were
not maintained even in the face of her emails to people, apparently seeking instructions.
For example, she was asked about an email she sent to Marie Cockett, Keith Baines, David
X Smith and Rod Ismay, amongst others, in which she asked for a decisionon settling the
claim.’ She was asked why she had chosen that group of people to send the email to, and
she recalled that Keith Baines had given a witness statement (presumably in the Cleveleys
case), but when asked whether it assisted her in recalling who gave her instructionsin the
Castleton case, she said that it was a whole selection of people, a moveable feast, and no
one from the list of addressees stood out * Her otherwise incomprehensible reluctance to
recall or confirm who was giving her instructions is explicable if the instructions she
received were not all in writing, and were best forgotten if she now wants to avoid an
investigation which may lead to a prosecution*’.
That is the background against which to set the litigation strategy deployed by POL. Even
before Morgan had fully developed his Shaw v Picton argument, he was seeking ways to
avoid the need to prove that the Horizon IT system was reliable, as can be seen from the
August 2006 Bond Pearce email summarising his advice:
A further point made by Richard Morgan was that we should endeavour to move the main
area of focus in the case away from the Horizon system if possible. Richard suggested a method
to do that would be to prove (if possible) the physical cash losses at the M arine Drive branch
by reference to all the other documentation created around the transactions, not simply
by reference to what was in fact recorded on the Horizon system?”
This advice from Morgan put the onus on POL to ensure that they did not run the
litigation in this way unless the Horizon system was sound, because at the same time he
asked them to confirm that it was. With costs already rising far above the value of the
claim, it would have been justifiable to avoid instructing experts to prove that Horizon
5 POL00082537 pp1-2
36 1NQ00000979 p15 (internal numbering p59)
37 POL00113909
38 1NQ00000979 p16 (internal number 62-63)
39 It should not be forgotten that the Castleton judgment, on what is now known, might well be set aside on
grounds of fraud in the Civil Courts
*° POL00082537 p1
SUBS0000027
was sound if Post Office had reason to believe that it was, but in fact, of course, the POL
group by now had every reason to believe (at the very least) that it was not.
42. This email from Bond Pearce to Mandy Talbot should have been the final warning to the
POL group that they were crossing a line. If they pursued the legal strategy being
advocated to them, the litigation against Mr Castleton would become an intentional
sleight of hand: their lawyers would seek to deflect attention from the Horizon system in
court, so its reliability would not in fact be tested, but meanwhile the litigation would be
pursued aggressively so that it would appear as if they were defending the Horizon
system, so as to deter other SPMs from challenging its integrity". Not only should it have
prompted Ms Talbot to reveal the Cleveleys Report to Bond Pearce; it should have led to
senior discussions to reinvigorate the recommendations following the December 2005
meeting; and the Castleton litigation should have been halted unless and until POL could
be sure that the debt they were claiming was, in fact, owed. Castleton is therefore a
watershed in this history of injustice.
43. In fact the POL group pursued the sleight of hand strategy, and they did so aggressively.
The emails between Talbot and the various addressees over the course of the litigation
makes this perfectly clear:
a. Initially, on 4 September, she forwarded the Bond Pearce email to Baines, Ward, Utting
and others, and she asked the three of them to give her“assistance” responding to the
enquiries about Fujitsu, which included the request for confirmation that there were
no issues with the system.” Whatever assistance they gave, it did not include
providing honest answers, such as informing Bond Pearce about AI376, or the
problems with ARQ data.
b. When Talbot wrote on 9 November 2006 seeking instructions on settlement, referred
to above, she said POL should aim to get judgment in the full amount because "we will
be able to use this to demonstrate to the network that despite his allegations about
Horizon we were able to recover the full amount from him. It will be of tremendous
use in convincing other postmasters to think twice about their allegations".** During
November, when POL believed that the case would settle, they worked up wording
which they wanted Mr Castleton to sign, saying that he was wrong to blame the
Horizon system, and the shortfalls were due to his own mistakes*4 Of course they had
no evidence to support the statement they wanted him to extract from him.
c. Then when judgment was handed down in POL’s favour, on 21 January 2007, Talbot
messaged a large group of addressees, saying "the judgement has entirely vindicated
the Horizon system". She said they would be awarded costs at the indemnity rate and
added, gratuitously, that "Mr Castleton appeared to be stunned by the result". Rod
Ismay responded "great news" and "what can we do ona proactive comms front here?"
so as to "assure branches and clients that they can rely on the integrity of Horizon" >
44. All this, despite what Morgan said the case was about when he opened it:
Judge - The biggest issue in this case seems to be whether the computer 6 working properly, isn't
it?
“1 Although not in evidence, the authors have seen letters sent to an SPM, citing the Castleton case as being
proof of the robustness of Horizon. These have been sent to CTI.
*2 POL00090437 p65
“3 POL00090437 p64
“4 Eg. POLO0090437 p59
“5 POLO0157980
10
SUBS0000027
Mr Morgan - Well that that's how Mr Castleton would like to portray it...
... much like a pocket calculator, a computer is only a tool that reflects the information that's
entered on to it.
... And your Lordship in fact touches the core of this question the core of this trial and that is, is
this a trial about an account produced by an agent... which is verified by him or is this a trial
which is a rampage through how a computer works...
Misconduct by POL’s lawyers
45. Itis clear that Talbot intended to run the Castleton litigation for a collateral purpose, with
the debt claim a side show to defending Horizon at all costs, including at the cost of
ruining Mr Castleton. Given the documents which were found and shown to her, she had
little choice but to admit this when giving evidence to the Inquiry.”
46. Stephen Dilley, on the other hand, accepted that POL wanted to defend Horizon, but
claimed that this was not the main purpose, and sought to deny obvious conclusions to
be drawn from the documents. He said “when the claim was first issued, it was issued to
pursue what Post Office believed was a debt. However, as the case continued, I think the
motivation of Post Office changed and what they wanted out of the case changed. I think
it was less about making an example of Mr Castleton and more about sending a message
that they were willing to defend the Fujitsu Horizon System.” When he was challenged
with the documents which made it clear that Talbot / POL wanted to make an example of
Castleton, he resorted to denying that the documents meant what they plainly said:
Q: ... it certainly seems, as far as Mandy Talbot was concerned, that the case shouldn't be settled
and that Mr Castleton should effectively be sacrificed in order to prevent further challenges
against the Horizon System. Do you agree with that?
No, I don't.”
47. Mr Morgan KC sought to give an account which was even less in keeping with the
contemporaneous documents, claiming that he did not even know that the Post Office
wanted to be seen to be defending Horizon. This passage is revealing:
Q. Was there any suggestion made to you at this time that it was important to the Post Office to
vindicate the reputation of Horizon?
A. No and, as I've just been saying, that didn't form part of my strategy nor was it communicated
to me, nor could I have run it, had it been -- had I been told to do it on the basis of the material
that I had. [Emphasis added]
48. This answer identifies why Morgan did not want to remember the truth: he did not have
the evidence to support the contention that Horizon was reliable, and so if proving that
contention was POL’s purpose in running the litigation, it was a collateral purpose,
tactically unworthy, and an abuse of the process of the court.
49, Having just made it clear why he should not have allowed POL to pursue their collateral
purpose, Mr Morgan KC was then taken to a telephone attendance note by Dilley, who
*©LCAS0000197 p14
47 A. ... it was indeed about the desire on the part of POL to have a substantial judgment dealing with
allegations about the Horizon System.
Q. That's what you delivered for them, wasn't it?
A. Yes.
https://www.postofficehorizoninquiry.org.uk/file/1837/download?token=y1IOqlijp177-178
#8 1NQ00001077 p4 (internal numbering p14)
49 1NQ00001077 p6 (internal numbering p21)
5° 1NQ00001078 p34 (internal numbering p134)
11
SUBS0000027
recorded himself as telling Morgan “the Post Office driver had been getting a judgment
against Mr Castleton to show that the computer system wasn't wrong and deter other
subpostmasters from bringing a claim”.®' Morgan accepted the obvious, but he went on
to try and extricate himself by saying that he may not have “focused” on this aspect:
Q. So what this note records you as having been told, a month or so out from trial, was that the
driver for the Post Office had been to get a judgment against Mr Castleton to show its computer
system wasn't wrong and to deter other subpostmasters from bringitg a claim. That's not about
recovering the money, is it?
A. No, Lagree.
Q. It wasn't about the sums involved in either the claim or the counterclaim?
A. Yeah, I can see that.
Q. And you were being told, according to this note, it was to get a judgment to show the integrity
of a computer system and about deterrents?
A. Yeah, I can see that.
I didn't recall this as being information conveyed to me. I'm quite surprised to see it there now.
I don't recall it at the time and had I focused on that, I think my response would have been that
I couldn't -- I simply couldn't prove that the system wasrit wrong. It just wasn't an achievable
objective.
50. If he did not focus upon it then he should have done.
51. This was not the only time Morgan failed to focus on an important issue. He told the
Inquiry that Castleton had not disputed the cash accounts in his testimony at trial. The
transcripts show this to be untrue: Castleton explained at trial that he had disputed the
accounts by calling the Helpline. When this was put to Morgan, he claimed that the
“impression” he had received was that Castleton accepted that the accounts were a fair
and true reflection of what had occurred in his branch. If this was genuinely his
impression, as opposed to a rosy memory to appease his conscience, he should have
listened to what Mr Castleton was actually saying, as opposed to what he wanted to hear.
52. We submit that both Mr Morgan KC and Mr Dilley should be referred to their respective
professional regulators for allowing POL to abuse the process of the Court by pursuing a
collateral purpose in this case. Running up costs of more than £321,000 was itself a clear
sign that POL was not really pursuing a debt of £24,000, and both lawyers had ample and
clear evidence of POL’s true intent.™
53. To make matters worse, Morgan drafted® and Dilley signed a Re-Amended Reply &
Defence to Counterclaim® which said:
Fujitsu Services have looked at the Claimant's computer system and have confirmed that the
losses recorded by the Defendant were caused by adifference between the physical transactions
that actually occurred and were recorded on the systemby the Defendant or his assistant as
taking place and the cash in hand that was declared by the Defendant relating to those
transactions, and accordingly those losses were not caused by theClaimant's system's software
or hardware.”
51 POLO0069794
52 1NQ00001078 p34 (internal numbering pp135-6)
53 1NQ00001078 p40 (internal numbering pp157-9)
4 PQL00070811 “there are important broader implications at stake such as the message it will send to other
subpostmasters if the [Post Office] settle or are seen to pursue it vigorously”
55 POL00069801 is an email from Morgan to Dilley which makes it plain Morgan drafted this Statement of Case.
5° Dated 23 October 2006. Mr Castleton was to become a litigant in person on20 November 2006
57 LCASO000190 p1 §3
12
SUBS0000027
54. One month after this was signed, Mr Castleton become a litigant in person, and,
unsurprisingly, he mounted no formal challenge to this paragraph. There are no
contemporaneous documents which suggest that there was an evidential foundation for
the claim made in it. When Dilley was asked about it, he suggested that it was based on
the evidence of Anne Chambers or Andrew Dunks,* neither of whom attested to looking
at the Horizon system and making the findings as drafted. Mr Morgan KC said the
paragraph was based on instructions from his solicitor but was not able to identify when
or how Dilley had given him that information.”
55. This took the legal sleight of hand one step further. The Statement of Case did not merely
side-step the question of Horizon’s reliability, it met it head on, with an assertion that was
not supported by evidence. It nevertheless bore a signed statement of truth from Dilley,
despite the fact that Morgan knew he did not have the “material” to “vindicate the
reputation of Horizon’.
56. Given the defence Mr Castleton put forward, and the way POL was responding to it,
Dilley and Morgan should have verified and documented the evidential foundation of
this paragraph very carefully and extensively before settling, signing and serving the
Statement of Case. The process of supporting such an assertion should, in fact, have been
a memorable part of the case for both of them.
57. There is nothing to suggest that Baines shared the knowledge of Horizon’s flaws with
Dilley and Morgan, unlike the internal POL group behaviour in the aftermath of the
December 2005 meeting. We do not, therefore, suggest that Dilley or Morgan intended to
pervert the course of justice. Nonetheless, serving this amended Statement of Case, in the
context of their client’s clear collateral purpose, was conduct which (we say) fell very far
below the standards of integrity expected of Officers of the Court.
The Fujitsu group conspiracy
58. The Inquiry is yet to hear from Pinder or Elliott, and so this section must be considered
with a degree of caution. What is inescapable, we say, is that Anne Chambers gave
perjured evidence in the Castleton trial. She would not have done so without higher
authority. The culture and motivations of Fujitsu management indicate culpability.
Knowledge within the Customer Services Directorate
59. Litigation support sat within this Directorate. We refer to §28 of our Phase 3 Closing
Statement, for a summary of the evidence of the Customer Services Director in post in
2000, Stephen Muchow. He said that he knew that the EPOSS code had too many bugs,
and as a result his department was the sacrificial lamb, by which he meant that SSC and
the other lines of Horizon support had to manage the fallout from the bugs.
60. The whistleblowing testimony of Richard Roll has now been accepted by every relevant
Fujitsu witness, including Muchow: SSC was routinely inserting transactions into branch
accounts in order to try to make them balance. This is, of course, not the sort of activity
the makers of accounting software should have to resort to, and it was, in itself, very clear
evidence that Horizon did not produce reliable cash accounts.
°8 1NQ00001078 p1 (internal numbering pp 3-5)
59 INQ00001078 p28 (internal numbering pp111-2)
®° 1NQ00001078 p34 (internal numbering pp135-6) as per footnote 54 above
13
61.
SUBS0000027
We submit that subsequent Customer Services Directors, such as Naomi Elliott who was
in post in 2006, must have been equally aware of the heavy load which was put on SSC
and Horizon support as a result of the poor code built into Horizon at the outset.
Fujitsu’s failure to understand the role of litigation support
62.
63.
65.
66.
We have heard from Paul Patterson that litigation support was not a service Fujitsu was
equipped to deal with:
A. I was professionally very surprised that that service even existed. We're meant to be an IT
company not a prosecution support service and, for that to be designed in from the very earliest
stages, I was very, very surprised at it. And in terms of the work associated with doing it, I have
no view on it. Tam amazed that it was even in the contract.
Q. Are you saying, in effect, that this is not something that Fujitsu was set up to do and not really
something that Fujitsu was skilled at or able to do properly?
A. ... there is no contract that I've ever signed where that obligation and extent of that obligation
sits inside it. I do not understand why it was there. And, clearly, it was there for -- and is still
there, actually -- for a very, very long time.
Q. ... we see the consequence, don't we, of that sort of exceptional, and perhaps unprepared for,
contractual service because, as we know, Anne Chambers gave evidence in less than ideal
circumstances in the Castleton trial.
A. lagree.
When giving these answers Patterson may have had in mind the appalling testimony of
the litigation support manager in post at the time of the Castleton trial, Peter Sewell.
According to Sewell’s own email, he thought the purpose of litigation support was to
send witnesses to court to oppose the likes of Castleton, who was “out to rubbish the FJ
name”.®? When faced with that email, his patently dissembling answers further
demonstrated how disastrously misplaced he was in managing a team of people who
were required to provide evidence for use in court.
. He even claimed that he did not know that any SPMs had complained of problems with
Horizon cash accounts. This was an obvious lie. Not only was he managing a team which
responded to SPM complaints about the integrity of Horizon accounts, but there is clear
documentary evidence that he was part of the Fujitsu response to Cleveleys in 2003,
during which he was told that Mrs Wolstenholme was complaining of the problems she
had experienced balancing.“
The fact that Sewell was in a role which should not have had the likes of him anywhere
near it was not merely unfortunate. It was part of the wider Fujitsu failure to perform on
the contract with POL, both in terms of the adequacy of the product, but then ultimately
in the attempt to make up for that through maintenance and support. By the time the
Castleton litigation was underway, there was already a very troubling history within
litigation support, which we will return to below when considering criminal
prosecutions.
The wider failure of the litigation support function is exemplified in John Simpkins’
admissions that the ARQ data was not sufficient to draw conclusions about the
51 NQ00001117 p29 (internal numbering p115-6)
© FUJ00154750
®3 1NQ00001116 pp32-22 (internal numbering pp126-9)
°4 He claimed to have no memory of this whenp220 of POLO0118221 was put to him - 1NQ00001116 pp31-2
(internal numbering 124-6)
14
SUBS0000027
functioning of the Horizon system, because access to the full message store was needed,
and further, that he and his colleagues had never been asked to advise on the kind of data
that should have been supplied. Even more specifically, he admitted that the Marine
Drive ARQ spreadsheets he looked at for the purpose of giving evidence at the Inquiry
did not provide sufficient information to assess the health of the Horizon system’s
performance, but that some of the information needed could be obtained in branch by the
SPM running off reports. We have taken instructions on this, and it was exactly these
reports which Cath Oglesby removed from the Marine Drive branch, after which they
were never seen again. Mr Castleton’s recollection is to be preferred against hers.
67. Whether that was by accident or design, the Fujitsu litigation support function was
evidently falling very far below the mark, which would have been exposed by the
Castleton litigation. This may account for Pinder’s initial responses when first contacted
about it, which seek to deflect, and require POL to send a standard criminal prosecution
request for ARQ data.°’ By that point, Fujitsu and POL had reached a mutually agreeable
arrangement for the provision of inadequate data in criminal cases, so Pinder clearly
hoped he would be able to do the same in the civil case.
Using Anne Chambers to avoid exposing Fujitsu’s failures
68. In the event, it was Dilley who prevented the complicit ARQ arrangement from working
in the Castleton litigation. Whatever his shortcomings, he was not immune to the need
for some kind of evidence to support a case, unlike Rob Wilson’s criminal law team.
69. Throughout the Autumn of 2005 he made a number of attempts to secure some kind of
evidence from Fujitsu in response to the reports that Castleton’s solicitors had sent to
him.* Eventually in December 2005 Pinder responded more substantively, via Ward.” It
is clear from this reply that he was dependent on Anne Chambers and Gareth Jenkins to
provide him with information.
70. Thereafter, throughout 2006, these two potential witnesses were continually in the offing.
There was a meeting on 6 June 2006 attended by both, as well as Elliott, Pinder, Sewell,
Ward, Dilley and others. The note makes it clear that Jenkins was senior toChambers, but
she had been directly involved in one of Castleton’s calls to the Helpdesk”
71. This was followed by them working together on a response to Castleton’s analysis of the
data from Week 42. The Chambers draft” was “firmed up” by Jenkins,” which again
shows that he was the more senior figure, and the one who was more comfortable
providing evidence that supported Horizon.
72. Nevertheless, it was Chambers alone who ultimately gave evidence. The reason for this
has not been recorded clearly in contemporaneous documents. There is a tangled email
chain which does not shed much light. The more illuminating evidence came from Mik
Peach in Phase 3, and we refer to §§49-51 of our Phase 3 Closing Statement for a summary
® 1NQ00001115 p11 (internal numbering pp42-4)
66 1NQ00001115 p13 (internal numbering pp50-52)
®7 POL00090437 p128-30
8 POL00090437 p101 onwards
°° POL00090437 p94
7° POLOOO71414
71 LCAS0001306
7” WBONO000027
73 POLO0081490_034
15
SUBS0000027
of the pressure that was brought to bear on Peach by Pinder and Elliott, obliging him to
offer Chambers up as a witness.
73. The Fujitsu group were conscious of
a. the unreliable Horizon cash accounts,
b. _ the regular need to insert transactions into branch accounts as a result
c. the problems that presented within litigation support, which was supposed to
provide evidence that POL could rely upon to prove that cash accounts were reliable
and SPMs were in charge of all the entries.
74. The initial reluctance to assist, and the later internal manoeuvres which resulted in Anne
Chambers giving evidence, suggests that the Fujitsu group were panicking, but
ultimately doing their utmost to prevent the High Court from hearing the truth about
Horizon’s frailties.
The response to Chambers’ Afterthoughts
75. Against this background, guilty knowledge can be inferred from the complete lack of
response from Pinder and Elliott to Anne Chambers’ Afterthoughts.4 As with the POL
response to the December 2005 meeting on Horizon integrity, there is no good reason for
the failure to act upon her sensible suggestions:
a. She pointed out the blurred lines between her evidence of fact about the Castleton
call and the eventual questions which led to her giving expert evidence about the
system as a whole; and she made the point that in future there should be a proper
technical review to support that kind of expert evidence.
b. She noted that the message store was only disclosed as an afterthought, that the
Tivoli event logs were only disclosed after she had revealed they existed when giving
evidence, and that there were other types of files archived to the audit servers, as well
as material attached to Peaks, all of which was potentially disclosable.
c. She even noted that it was unfair that SPMs were continually batted back and forth
between the NBSC and the Horizon Helpdesk.
76. Pinder's only known response, copied to Elliott, bears quoting in full:
Thanks Mik, there was no intention to have a wash up on this particular case as such but I must
stress that from the outset this was ‘new ground’ and a particularly unusual case (1st of its kind
in 10yrs) for all concerned. It involved many different variables which, at any point in time could
have culminated in a totally different outcome.
This enquiry took well over a year to conclude and routine procedures which have served us
well for 10 years were suddenly being stretched to new limits, but it does highlight how (POA)
can be called to account and I totally agree we must learn from this.
Ann (many thanks for your comments) you have highlighted some interesting areas of procedure
which we need to recognise, and I will discuss these with Naomi and will keep you both
informed.
77. Itis not merely a “pat on the head”. The full subtext is “phew, we got away with it” and
nothing was acted on. It appears nothing was implemented.
Anne Chambers’ perjury
78. Anne Chambers is plainly an intelligent woman. She had the sense to send the
Afterthoughts document, which — to some extent — covers her back. However she made
74 FUJ00152299
16
79.
80.
81.
SUBS0000027
one ‘slip’ in her evidence to the Inquiry. She said, unbidden, that she was told not to
mention the Horizon Known Error Logs when preparing for the Castleton trial. When she
was picked up on this and questioned closely, she said she would have received this
instruction from Peach, Pinder, or Elliott. She claimed not to realise that this was strange.”
To someone of Mrs Chambers’ intelligence, it must have been obvious why Pinder and
Elliott would not want the Court or Mr Castleton to know that the KELs existed - their
very name was a problem for Fujitsu (although it would not have been, were the known
errors nothing more than the usual run of the mill problems any system might encounter).
Fraser J found that they were essential for analysing the reliability of Horizon accounts. It
follows that hiding the KELs from Mr Castleton hobbled his ability to advance his case.
Being told not to mention essential evidence, and then abiding by that instruction,
indicates that Mrs Chambers did not tell the whole truth when testifying at the Castleton
trial. Her claim that she did not find this suspicious is not credible” It is also telling that
she did not mention the KELs in her Afterthoughts. She claimed that this was no more
than an oversight, but she did mention PEAKs,” which were something of a companion
document within SSC practice and procedure.
She also failed to mention in her trial evidence that she had conducted an investigation
into Marine Drive’s Week 42 data and found a missing stamp transaction.” As the
Castleton trial proceeded, it evidently became clear to Chambers that she was being asked
to give evidence about Horizon more generally, not just the investigation she carried out
as a result of Castleton’s call to the Helpdesk. She knew that Castkton thought that there
were missing transactions in the Week 42 data, because that was why she had been asked
to look at it. When he started asking her questions about problems in the Horizon system,
answers which told the whole truth would surely have included an explanation of the
missing transaction, and the “known fault” which had caused it. That in itself would have
revealed the existence of the KELs, since that is where the known fault must surely have
been recorded. She failed, in all respects, to do so. These omissions amount to deceit.
It is also notable that in her evidence to the Inquiry, although Mrs Chambers has been
forthcoming, she has also been reluctant, at times, to provide a simple, true answer with
no bark on it. We submit this is because if she had it would have presented a stark contrast
to her testimony in the Castleton trial, and draw attention to other problem areas for
Fujitsu which she chose to gloss over. It took Counsel to the Inquiry several attempts,
over the four days of her evidence, to extract what was, eventually, an admission that SSC
staff could and did use SPM log in details to insert transactions, such that it was not
possible to be sure that the transactions which appear on the Marine Drive ARQ data
actually took place in branch.” Again, as an intelligent woman, she must have known at
the time of the Castleton trial, that this was information the Court should know.
75 1NQ00000980 p12-13 (internal numbering pp45-49)
76 Q. Did it not strike you even then as perhaps slightly suspicious that nobody wanted you to mention known
error logs with that title being what it was?
A. I don't think I thought of it as suspicious. I thought it seemed strange but, as I said, I was in a very unfamiliar
situation. https://www.postofficehorizoninquiry.org.uk/file/1824/download?token=qsAmi2y p147
77 1NQ00000980 p37 (internal numbering 145-147)
78 WBONO000027
79 Q. So transactions which appeared in the standard filtered ARQ data, for example, in Mr Castleton's case,
with his ID user number next to them, would not necessarily mean that they were carried out by him?
A. It would have been possible, yes, for SSC to put transactions in, that--
Q. Using his ID?
17
SUBS0000027
82. There were repeated examples of Chambers seeking to appear even-handed before the
Inquiry, when her answers were really aimed at shoring up her original testimony. The
evidence was often technical, and complicated, but one example is perhaps illustrative. It
was acknowledged by Chambers in the June 2006 meeting that hardware failures could
cause lost transactions.*’ On 2 November 2006, Greg Booth, the second temporary SPM
who came into Marine Drive after Castleton was suspended, informed POL lawyers that
he had experienced lost transactions after screen freezes. Pinder told Dilley that there
would be evidence of this in the ARQ data, in the form of “unusual’ restarts, and he said
that this was being checked.*! However, there was no evidence that these checks were
completed, and no one testified to it. Chambers said in her Inquiry testimony that she
could not remember if it was her who made the checks. In the event it seems that no one
followed up on this, because Pinder sought help from Jenkins to close the issue downas
user error, on the basis that Booth had not followed the Horizon User Manual.*?
83. After a tortuous passage of technical questioning, Ms Chambers tried to play down the
significance of this loose end:
Q. Well, just standing back and taking the view overall, if we may. Whatever examination there
was for unusual restarts in Marine Drive we now don't really have any conclusive evidence of it;
is that right?
A. There's -- [ haven't been able to find any evidence that it was happening but it is a possibility
that there were some and no, we don't have anything that would be conclusive.
Q. All right. What that could have shown is whether there were problems with screen freezes
and, therefore, potentially missing data following screen freezes or it would have been easier to
find the possibility, if you'd found unusual restarts; is that rght?
A. Yes. But, as I said, it wouldn't necessarily cause missing data, but it might —
Q. But it might do.
A. Potentially, there might have been sessions that didn't settle but that wouldn't necessarily
cause discrepancies.
Q. No, but it might do?
A. Unlikely to but, yeah, it would depend on the individual circumstances. *°
84. A genuinely even-handed witness would not have felt the need to temper every
concession, but a witness who is deeply conscious that she did not tell the whole truth
about problems in the Horizon system would want to suggest that any inadequacies in
the Marine Drive investigation were minor, and they did not make any real difference.
85. In the same vein, her explanation for the zero-stamp declaration, offered apparently off
the cuff, was a theory that sought to explain away an apparent anomaly. It relied on the
postulation that Mr Castleton had used “a different drawer ID”. However, we are
instructed that he did not have a different drawer ID, and this really goes to show how
Chambers was constantly casting about for theories — even if they were totally
unsupported — that would allow her to stand by her original testimony.
A. Using his ID.
Q. Without leaving a fingerprint on the standard, filtered ARQ data that that had been done?
A. Yes, I think that would have been possible.
://www.postofficehorizoninquiry.org.uk/file/1818/download?token=RB4Y3yRdp68
°° POLO0071165 p6
1 POLO0081826_022 p7
®2 See Booth’s Second Witness Statement, in which he blames himself for not following the manual:
LCAS0000471
3 1NQ00000980 p35 (internal number p137)
®41NQ00000980 p31 (internal numbering pp121-22)
18
86.
87.
88.
89.
SUBS0000027
She was even more keen to “cover off” the questions which Mr Castleton had put to her
at trial about his user ID apparently switching between Nodes 1 and Node 2. She
categorically rejected the suggestion that this was anomalous, despite her much more
equivocal evidence at trial.*° She explained the change on the grounds she had not had a
chance to think about it at trial,5* but the rock-solid explanation she wanted to rely on in
her evidence to the Inquiry was obviously self-serving. If Mr Castleton had exposed an
anomaly at the time of the trial it would undermine her assertion in her witness statement
that “there was no evidence whatsoever of any system problem”.*”
We submit that Mrs Chambers’ explanation for the Nodes switching should not be
accepted without independent expert evidence® supporting it. An authoritative textbook
on distributed systems (such as Legacy Horizon) says this:
A node in the network cannot know anything for sure—it can only make guesses based on the
messages it receives (or doesn’t receive) via the network. A node can only find out what state
another node is in (what data it has stored, whether it is correctly funtioning, etc.) by exchanging
messages with it. If a remote node doesn’t respond, there is no way of knowing what state it is
in, because problems in the network cannot reliably be distinguished from problems at a node.
Discussions of these systems border on the philosophical: What do we know to be true or false
in our system? How sure can we be of that knowledge, if the mechanisms for perception and
measurement are unreliable?
This suggests that it may be much harder to give categorical explanations for the
communications between Nodes than Mrs Chambers’ evidence would suggest.
It is unfortunate for Mrs Chambers that she was put into the firing line by Pinder and
Elliott, but we submit that she was the tool that they used to pervert the course of justice,
and against her own better judgement, she did not tell the whole truth in the Castleton
trial.
Stephen Dilley’s failures of disclosure and the drafting of Statements”
90.
91.
For the detail underpinning this section we refer to our Submission regarding additional
questions for Stephen Dilley, dated 30 October 2023, and the earlier additional questions
which we appended to it dated 22 September. In broad terms, however, we submit that
Dilley should be further referred to the SRA for 1) failures to disclose and 2) drafting
misleading witness statements.
The joint report by Chambers and Jenkins”! referred to a missing stamp transaction in the
Marine Drive ARQ data for week 42. This was prepared in response to Castleton’s own
analysis of the week 42 data, which asserted that there were missing transactions, so the
Report undeniably dealt with an issue in dispute in the proceedings and would have
assisted the defence. Dilley did not disclose it. In his Inquiry evidence he claimed to have
mentioned it to Castleton’s then lawyer in a phone call, but his own phone note of the call
about the week 42 data did not refer to the Report, or any discussion about it.” It is clear
85 1NQ00000980 pp35-6 (internal number pp138-143)
86 1NQ00000980 p32 (internal numbering p126)
87 LCAS0001265 pS
88 Mr Cippione, perhaps?
89 Designing Data Intensive Applications by Martin Kleppmann (O'Reilly Media Inc, March 2017) - Chapter 8
‘The Trouble With Distributed Systems’, sub-heading ‘Knowledge, Truth & Lies’
°° We suggest that a potential pattern emerges in POL’s civil litigation —e.g., the manner in which Elaine
Cottam’s (née Tagg’s) statement was taken, if she is to be believed.
° WBONO000027
°? POLO0069604
19
SUBS0000027
from the note that Dilley cherry-picked the content from the Report which helped his
client’s case, and presented that in the phone call as neutral information, without
attributing it. There is an argument that the Report was privileged, but there are
arguments both ways, and there is no evidence that Dilley ever applied his mind to the
question, let alone to the question of whether he waived privilege by using the helpful
material in the report when talking to the defence.
92. In any event, the material referred to within the report was disclosable on any view:
a. acomplete extract of audit data for the period concerned
b. areport which apparently would have been generated overnight when the missing
transaction occurred (although that was said to be no longer available)
c. the Reference Data in use at all branches at that time
93. Likewise, the reference to the “known fault” should have been provided, because that
was clearly relevant, and would have revealed the existence of the Known Error Logs.*
94. His failure with regard to drafting witness statements is also set out in our 30 October
Submission, but in a nutshell, it is clear from the contemporaneous email referred to
therein™ that Dilley was very alive to the content of the Exhibits to Helen Rose’s Witness
Statement. Under those circumstances, it should not have taken him long to realise that
they undermined the content of the statement. It was at best incompetent and at worse
malign to allow Rose to sign off on a grossly misleading witness statement.
95. The non-disclosure of the BDO Stoy Hayward Report falls into a different category, in
our submission, because there was a much clearer argument that it was privileged.
Nevertheless, it also disclosed problems, as had the Greg Booth missing transactions, as
had the huge number of calls to the Helpdesk, which again, Dilley, chose not to disclose?
Taken together, these were red flags requiring Dilley to be much more careful with
disclosure, and if he thought privilege applied to either of the reports he did not disclose,
he should have been clear about that at the time. It is our submission that whilst non-
disclosure of the BDO Stoy Hayward report is defensible as a commercial litigant, it raises
questions as to how Quasi-Departmental Public Corporations ought to act in litigation. A
higher ethical standard is required.
The criminal prosecutions conspiracy
96. Criminal prosecutions were conducted by POL in furtherance of its ruthless commercial
and reputational imperatives, contrary to the Rule of Law and the interests of Justice. POL
routinely failed to adhere, and disregarded guardrails prescribed by Law, such as PACE
Code C and the CPIA. We submit that there is evidence to support the following
allegation:
Rob Wilson, Tony Marsh, Tony Utting, Graham Ward, John Scott, David Pardoe, David
Posnett, Stephen Bradshaw, Ged Harbinson, Jan Holmes, Andy Dunks, Penny Thomas,
Brian Pinder, Peter Sewell (the Security group) conspired together, and with others, to
pervert the course of justice by pursuing criminal prosecutions between 2000 and 2015
which were not in the interests of justice, but which were in the interests of POLand
Fujitsu.
°3 The document which was not put to Dilley on this issue was his email to Morgan- POLO0069797
°* POLO0070790
°5 1NQ00001077 pp23-5 (internal numbering pp91-8)
20
SUBS0000027
97. POL’s deliberately inadequate investigations were a course of conduct which had a
tendency to pervert the course of justice, because they routinely i) relied upon evidence
from Fujitsu it knew to be misleading ii) failed to disclose material which met the
disclosure test / failed to follow reasonable lines of enquiry, iii) over-charged without
evidence and iv) prompted unreliable confessions, which allowed POL to secure
convictions and subsequent Confiscation Orders on a reversed burden of proof. This not
only unjustly enriched POL directly, but it also secured unjust commercial advantage for
POL and Fujitsu, by supressing the truth about POL’s unreliable IT systems. Disclosure
was wilfully degraded, ignored or discounted. This was mirrored by Fujitsu’s
repudiation of quality standards.”
98. We submit that there is evidence that this course of conduct was also intended to pervert
the course of justice, because the Security group either knew that POL’s IT systems were
unreliable, or they closed their eyes to that knowledge, so the failures in POL’s
investigations were either deliberate or deliberately overlooked. Key Fujitsu figures were
complicit in this. As stated above, the motive for the course of conduct may have been to
protect POL/Fujitsu, or their own jobs, and the natural consequence of the course of
conduct may not have been desired as such, but that does not mean it was not intended.
Perverting the course of justice may have been seen as unfortunate collateral damage.
The ad hoc contractual arrangements for POL prosecutions
99. When the Benefits Agency abandoned Horizon, POL had to negotiate how the new
system was going to evidence benefits fraud as well as audit shortfall prosecutions. This
was problematic, because, as AI376 shows, POL knew that the Horizon cash accounts
were unreliable. Cash accounts were central to the criminal investigation of benefits
frauds as well as audit shortfalls, as Rob Wilson explained a few years later:
Whilst the prosecution rely on the checks of the actual orders dispatched to Lisahally to prove
the overclaims themselves, the cash accounts remain a significant piece of evidence. The fact that
the overclaimsare deliberate, as opposed to accidental is demonstrated by the lack of a
corresponding surplus in the cash account?*
100.On 15 July 1999, David Miller, the Horizon Programme Director, agreed Terms of
Reference for a Review of the Horizon Cash Accounts.” On 20 July he is recorded as
telling the POL Board that he “considered the system robust, and fit for purpose”, yet
when he testified to the Inquiry, he said that if he made this statement it was incorrect,
and he pointed out that over the page it was recorded that “members were concerned
that a number of technical issues remained unresolved” 1°
101. This Review of the Horizon Cash Accounts (in an undated draft produced by Jeremy
Folkes from his own record) concludes:
There is a need to ensure that the problems relating to the audit trail for S&IE investigations and
demonstrating that the system meets the requirements of the Police and Criminal Evidence Act
have been impact assessed as incidents and are considered by the Acceptance and Release
°8 See Cross-examination of Pardoe by Henry KC re the culture of nor-disclosure - 1NQ00001100 pp39-41
(internal numbering 153-163)
°7 FUJO0155476 — Guy Wilkerson advises Peter Sewell to remove the phrase "fit for purpose" in relation to a
line on the admissibility of evidence in a CP (this amendment was made here: FUJ00155474)
°8 POLO0104593 p7
°9 WITN05970134 p1
409 PQLO0000352 p11, see also §24 of our Phase 2 Closing Statement
21
SUBS0000027
Authorisation Boards if not satisfactorily resolved. In addition, it will be necessary to consider
whether the current level of cash account errors will affect the accuracy of settlement with clients,
when considering the rate at which the system should rdll-out.!"!
102. Whatever action was taken arising from this, the problems with audit trials were
inadequately impact assessed. When Horizon rolled out, arrangements to obtain evidence
to support prosecutions were woefully inadequate. Even if ARQ data was obtained, it
was not sufficient to assess the health of the Horizon system (see §§66-67 above).
103. Furthermore, over Phase 4 it has become clear that many of the investigators we heard
from in Phase 4 were recruited in 1999/2000 as Horizon was being rolled out. This was a
recruitment drive which ran alongside the negotiations around AI376. POL’s senior
management should have known that unless these new investigators were very well
aware of the potential for cash accounts to be unreliable, as well as equipping them with
the tools they needed to analyse whether they were reliable, grotesque injustices would
occur. Phase 5 will need to examine who was responsible for this recruitment drive, and
whether there was a deliberate policy of suppression or disinformation.
Jan Holmes’ role in Cleveleys and the prosecution of Tracy Felstead
104. Jan Holmes was an influential person at this early stage. He was responsible for providing
ARQ data to POL before that function moved from Audit to Customer Services. Then in
2001 Mr Holmes was involved with the ARQ missing data problem, referred to at §§17-
18 above. Then in 2004 he was instrumental in the Fujitsu response to the Cleveleys
Report from Jason Coyne, and was remarkably bullish in his defence of Fujitsu.”
105. This was despite the fact that he co-authored the 1998 Report on EPOSS PinICL Task
Force" with David McDonnell, and then authored and owned the consequent Schedule
of Corrective Actions. This troubling document stated on 17 November 1999 that “EPOSS
continues to be unstable”, but in the following months Mr Holmes reported the various
ways which Terry Austin claimed to have dealt with the problems without following the
original recommendation to re-write the EPOSS code. These claims did not bear scrutiny,
as Mr Holmes himself saw, because he attempted to obtain statistical evidence of whether
the fixes had worked, but he was stonewalled. On 10 May 2000, Mr Holmes recorded the
decision taken by Mike Coombs to close the issue down.
106. Extraordinarily, this man, who was intimately acquainted with the flaws in the Horizon
system", and who was not a technician,’ provided a draft Witness Statement for the
Tracy Felstead prosecution which said this:
There are no reasonable grounds for believing that the information stored on the Horizon system
would be inaccurate because of improper use of computer terminal. During the Material Time
the Horizon system was operating properly at the Camberwell Green Pos Office Outlet or if not,
any respect in which it was not operating properly or was out of operation was not such as to
affect the production of audit records or accuracy of their contents!”
40 WITNO5970134 p5S
122 See, for example, this section of his evidence:INQ00001073 pp18-19 (internal numbering 71-74)
403 FUJ00121098
104 WWITN04600104 p9-10
105 He even reviewed FUJO0121098 on 14 May 2001 but was not forthcoming as to why he did so
106 https://www.postofficehorizoninquiry.org.uk/file/855/download?token=IqWbSNnip107, line 23
107 WITN04600217 p5 There is no doubt this statement was used against Miss Felstead— see §59
22
SUBS0000027
107. Even more extraordinarily, Fujitsu sought to charge £19,047 for providing it to POL, and
POL sought to defray that by making the defence pay for it. We refer to our Note to the
Inquiry relating to Jan Holmes, dated 17 August 2023, for the detail, and the remaining
questions arising from this troubling evidence. The result was the burden of proof was
reversed, the defence did not have the material they needed to discharge the burden
wrongly placed upon them, and an innocent 19-year-old girl was sent to prison and
nearly sent out of her mind.
108. We say the decisions around the significant sum of money that Fujitsu were seeking as
payment for providing the Holmes statement must have been escalated within bothPOL
and Fujitsu; and it is known that senior people were taking the decisions with respect to
Cleveleys (see §19 above).
Symbiosis: Fujitsu (Holmes) and POL
109. While Holmes was helping POL to cover-up Horizon problems in the contested and
therefore open cases of Cleveleys and Felstead, POL security was operating more covertly
in “dark places”,’°* as Tony Utting put it in his Inquiry testimony. In December 2005,
Graham Ward said this when referring to the Holmes Witness Statement, when
proferring it for the Castleton case, as an exemplar:
I would suspect that the Jan Holmes statement is more or less exactly what you'll need should
the ‘Castleton’ case proceed all the way (however I seem to recall that at the time, as it was out of
the normal this statement did costus 'an arm and a leg’.... but I maybe wrong).
110. What that tells us is that by 2005 a great many people had been prosecuted on the back of
evidence of this ilk, and these other cases were carrying on invisibly, below the radar.
Cases like that of Parmod Kalia, sent to prison after he pleaded guilty at the magistrates’
court on 17 January 2001.
- The conclusion which we say can be drawn at this stage is that those in charge of POL
investigations, such as Tony Marsh and Tony Utting, but more importantly the directors
responsible for their activities, closed their eyes to the known problems with Horizon cash
accounts, and this was despite the very obvious, visible cases of Felstead and Cleveleys
which were staring them in the face.
11
Designing the litigation support service
112. Those cases which were not ‘out of the normal’ were evidenced according to thebusiness-
as-usual litigation support practices which were developed by Fujitsu, with input from
POL. In February 2002 these practices were formalised for the first time. Prior to that,
there had been an “informal agreement” for the provision of up to 50 ARQ extractions
per annum.!"” The people involved in formalising the arrangements included Jan Holmes,
and the Approval Authorities included Martin Riddell, Customer Services Director, and
108 He was explaining why the Security Team Monthly Report started in 2004, because it was thought their
activities needed to be more visible to the top level of the business:INQ00001096 pp8-9 (internal
numbering 32-35)
209 PQL00083161_009 p1 - which reveals Holmes’ statement was deployed against a counter-clerk at
Camberwell, i.e., Ms Felstead [“...please find attached the statement from Jan Holmes which was
used in a prosecution of a counter clerk at Camberwell Branch Office in 2002.”]
420 FUJ00152189 p7
23
SUBS0000027
Graham Ward at POL." The policy that was produced is notable for the fact that the
number of per annum ARQ extractions leaped to 500/12 and because it produced a
flowchart for the ARQ extraction process and the litigation support process generally.!>
113. Both of these were subject to later change. A hard copy file of printed emails marked
“Audit Record Requests (Increase in Limits)” has been disclosed, which shows that from
2003 to 2008 there were numerous high-level decisions both to reduce and to increase the
limits. This obsession with the quantity of ARQ requests, and the cost of them, substituted
for any consideration of the quality of the prosecution support service being offered. At
the most fundamental level, it was not fit for purpose, as Simpkins’ made clear when
saying that no one in SSC was consulted to make sure that the information provided did
what it was supposed to do (see §66 above).
114. Worse than that, however, it failed to ensure that Fujitsu disclosed the material which
POL was obliged to consider for disclosure to the defence. In the 2002 flowchart, it was
promised that litigation support would include not only checking Helpdesk logs and non-
polling reports, but also “Analysis all event and fault logs”, and this was to take place
before completing the witness statement to produce the ARQ data. That was further
explicated at paragraph 7.2.3 as follows:
Any relevant PinICLs identified in Powerhelp logs will be reviewed through PinICL Client to
ensure that any recorded faults, would not hinder the outlets performance or otherwise affect
the integrity of audit archive from which the Record Queries are extracted. ThePinICL log will
detail the error relating to the site, equipment and or service in question. !*
115. By 2007, there was a revised Policy document, which also involved Ward, but on the
Fujitsu side the contributors included Penny Thomas and Dunks, with Sewell and Elliott
as Approval Authorities."° The revised flowchart had watered down the analysis of fault
logs to say merely “Check appropriate PEAK logs, if required”,!"6 and 7.2.3 had also been
watered down to say this:
If requested, all relevant Powerhelp calls will be reviewed to identify any recorded faults,
that might affect the integrity or admissibility of the audit archive from which the ARQs
are extracted.!!”
116. It was therefore very hit and miss whether PinICLs and PEAKs would even be considered
by Fujitsu, let alone disclosed to POL. There was reference to the complete message store
as the source of the data being extracted into the spreadsheet which was ultimately served
on disc, but no consideration was given to when it would be appropriate to disclose
further information from the message store to POL or the defence.
117. Neither document mentioned the Known Error Logs. This adds to the evidence that
Elliott and her predecessors and successors in the role of Customer Services Director were
deliberately hiding the existence of the KELs. Both documents had a section about
“Additional Litigation Support” which set out the ways that expert evidence might assist,
including by the provision of the Tivoli and other “system security event files”, but this
11 FYJ00152189 p1
12 FYJ00152189 p9
13 FUJO0152189 p16
14 FYJ00152189 p18
45 FUJO0122366 pl
46 FUJ00122366 p16
17 FUJ00122366 p19
24
SUBS0000027
disclosure was expressly excluded from the standard service, and it was envisaged to be
wholly exceptional.!"6
118. Worst of all, the documents hid Fujitsu’s “remote access” capabilities. The draft pro forma
witness statement appended to the 2002 policy had a long section explaining the way that
transactions in branch were recorded to the Transaction Management Service. It
contained this line:
This creates a record of all original outlet transaction details including its origin - outlet and
counter, when it happened, who caused it to happen and the outcome!”
119. This was not true. It failed to reflect the fact that some of the transactions in ARQ data
may have looked as if they were “original outlet transactions” when in fact they were
inserted by Fujitsu SSC staff. A search on CP View shows that this line appeared in some
270 Witness Statements between 2 March 2001 and 11 November 2008.
120. This was a knowing lie. Both policy documents included extensive sections explaining
the secure process laid down for extracting ARQ data, making it clear that Fujitsu
understood the importance of maintaining complete data integrity in the context of
prosecutions. Indeed, the whole thrust of the pro forma witness statement was to assert
that the exhibited ARQ data was a reflection of branch transactions, without any entries
added later.
121. The lie was constructed by Fujitsu, but at some point, during this period, POL came to
know it was a lie. As Mr Atkinson KC has shown, the fundamental importance of third
party disclosure and the Attorney General’s Guidelines on Disclosure were marginalised,
seemingly part of the covenant of secrecy between POL and Fujitsu to prevent this fiction
being exploded before the courts.
Remote access
122. From the inception of Horizon, there was a fundamental security flaw, which allowed
SSC staff to routinely tamper with branch accounts using “remote access” to take over a
SPM’s user ID, and then insert transactions which would look indistinguishable from
branch transactions on the ARQ data’°. Leaving aside all the other problems with
Horizon, this alone undermined every single prosecution based on Horizon data. Rob
Wilson conceded this, whilst denying that he knew of remote access in 2010, claiming
absurdly that he thought the Receipts/Payments mismatch issue notes related to visits by
Fujitsu to perform covert corrections on the physical counter at the branch."”!
123. The fact that this information was kept secret until 2018, speaks to how significant and
damaging it was, and how obvious that was to anyone who came to know about it earlier.
Very concerted efforts must have been made to supress that information over sucha long
time. Even now, all the SSC witnesses apart from Richard Roll have sought to downplay
it, by claiming that it was done infrequently, and it was done with POL’s knowledge.
Their motivation is clear: what they were doing was so obviously wrong they can only
excuse themselves by minimising and deflecting. However, they did not know about the
lie in the standard Witness Statement used to serve the ARQ data.
428 FUJO0152189 p22 and FUJOO122366 p22
119 FYJ00152189 p30
120 FJ00088036
221 Cross examination by Henry KCon 12 December 2023 re FUJ00081584 “This solution could have moral
implications of Post Office changing branch data without informing the branch’ from p 172/20 or 43/75
25
SUBS0000027
124. Andy Dunks was the main liaison between SSC and litigation support, but the chain of
command above him led to the Customer Services Director, who was responsible for the
SSC, so the various people in that role must have known about both sides of the equation.
125. Meanwhile, by October 2008 Post Office were well aware, of SSC’s capabilities, as
revealed in Andy Winn’s email to Alan Lusher about the case against the SPM Graham
Ward (as opposed to the POL employee Graham Ward).'? In 2019 Matthew Lenton
asserted that Post Office approval was sought whenever SSC inserted transactions which
would have an impact on financial data, and the person approving would be a “senior
Post Office manager” .!?>
126. We anticipate that the next two Phases will bring forth more evidence of the senior people
involved in conspiring to hide Fujitsu’s “remote access” capabilities. This is likely to have
its origins in the disappearing email from Lynn Hobbs to Rod Ismay in November 2010,
in which she explicitly told him about SSC’s capabilities. This also went around other
members of the POL senior leadership team including Angela van den Bogerd. We refer
to §§72-75 of our Phase 3 Closing Statement for the details.!*4
127. However, the cover-up which began with the Ismay Report and continued through to the
Bates litigation was, we submit, a different nature of conspiracy. The course of conduct
which formed the basis of the criminal prosecutions’ conspiracy involved relying upon
ARQ data served under knowingly misleading witness statements.
The 2008 ARQ problems
128. In the latter years of Legacy Horizon, Fujitsu became aware of significant problems with
the ARQ data which are explained by Patterson in §§75-116 of his Third Statement.'> In
very short form, these were locking problems which occur within distributed systems like
Legacy Horizon: sometimes Nodes need to lock, to prevent multiple Nodes from
simultaneously recording incompatible information; but equally, when a Node locks, it
can cause processes to fail. During the internal Fujitsu discussions, Gerald Barnes
identified what he believes was the more fundamental problem: poor error handling. On
2 January 2008 he said:
The fact that EPOSS code is not resilient to errors is endemic. There seems little point 'fixing it in
this' one particular case because there will be many others to catch you out...
It may be worth passing on the general message to the HNGx team that in many cases code
should always try and exit gracefully after an error and not just blunder on regardless.
This is a perfect example of why. Had the balancing code exited gracefully then if the user had
tried again after CABSProcess had finished working then all would have been well!?°
129.These locking problems combined with poor error handling resulted in lengthy
discussions about whether to fix, what kind of fix, what kind of checks to instigate, etc. It
was another year before they finally decided to inform POL. On 7 January 2009, Wendy
Warham, Operations Director, sent an email to Sue Lowther and David Gray of POL”’ in
which she claimed the problem had been “resolved”, and significantly downplayed what
had occurred, when compared with what is now known from Patterson’s Third
222 PQL00023432
223 FUJ00211295 p3
224 5450000022
225 wITN06650300
2 FYJ00155366 p3
227 POLO0142363
26
SUBS0000027
Statement. Peter Sewell was also downplaying the problem, as can be seen from an email
forwarding the Warham email within POL.'* Both, to a degree, acted disingenuously.
130. This, however, does not excuse POL’s reaction. David Posnett and Rob Wilson both took
the view that the standard witness statement from Fujitsu should not be amended. Rob
Wilson went on to say that there was nothing “undermining” and disclosure of the
information would “only lead to fishing expeditions”? This was fed back to Wendy
Warham and Peter Sewell by Penny Thomas, which made them complicit in POL’s
decision to hide the evidence of problems with the ARQ data.'** No doubt they would
claim it was not for them to decide how POL should run its prosecutions, but they were
responsible for their staff members providing witness statements to support those
prosecutions, and therefore the responsibility fell on them to ensure those statements
were truthful. Moreover, the wording they had proposed, and which Wilson and Posnett
rejected,!*' came nowhere close to revealing the whole truth as they all knew it to be.
POL Security was drenched in information about Horizon errors
131.The somnambulant response of Wilson and Posnett to the 2008 ARQ fault is also
revealing. By this point they had hundreds of prosecutions which depended on ARQ data
either completed or underway. If they had truly believed that Horizon was “robust” they
would have been shocked and horrified to hear that there was a problem that may require
an amendment to the standard Fujitsu statement exhibiting the ARQ data. The truth is
that by 2008 they were determinedly closing their eyes and ears to anything they did not
want to hear about Horizon problems, because in fact the evidence of the Horizon and
NBSC call logs was clear enough. Across all those investigations, there was ample
evidence of desperate SPMs calling in when they struggled to balance. It was even
accepted by Stephen Bradshaw that they were “drenched” in information “that Horizon
wasn't working” via these call records “from the very beginning.”"°?
132. The attitude of the security department to those call records was expressed in black and
white in April 2002 by Raymond Grant, a man who was so conditioned to that attitude
that he failed to even see the problem:
Miss Saleem indicates that the problem would appear to be" gletches in the system" she indicates
that she was told this by the helpdesk. It has not been possible to identify who, from the helpdesk
is giving out this information. It doeshowever give concern to Post Office Security that the
operators are being advised thatthe Horizon system is faulty and produces inaccurate results!
133. Evidently, he thought it important to try to prevent call handlers from telling SPMs the
truth about “gletches”, but he did not think Post Office Security would have concerns
about the inaccurate results they were causing. This reflects a culture not merely of
adversarial astigmatism, but myopia in respect of disclosure.
134. By the time Wilson and Posnett told Fujitsu not to disclose the 2008 ARQ problems, they
were already entrenched in intentionally perverting the course of justice.
8 pOLO0142363
229 FUJO0155400 pp2-3
230 FYJ00155400 pl
131 FYJ00155400 p3
132 1NQ00001112 p48 (internal numbering pp191-92)
+33 pOLO0093246 p1
27
SUBS0000027
Doctoring the Noel Thomas expert statement
135. Further evidence that Fujitsu and POL were working together to pervert the course of
justice comes from an email dated 10 March 2006 from Ward to Pinder, Thomas, Neneh
Lowther and copied to Sewell. It made requests for evidence in relation to Castleton and
other civil cases, and began the process for obtaining evidence in the Noel Thomas
prosecution. Ward said this:
given the allegations being made by the Postmasters, I'm sureyou'll agree that it is very much in
both ourselves and Fujitsu's intereststo challenge the allegations and provide evidence that the
system is not to blame for the losses being reported
136. This is a rare example of a conspirator putting the aim of perverting the course of justice
into writing. It is notable that none of the addressees, despite being in the business of
litigation support, took him to task, or escalated any concerns that a POL investigator was
nakedly admitting that he wanted them to provide evidence as a means of supporting
POL and Fujitsu’s commercial interests through litigation.
137. Small wonder that, as a result, the Jenkins statement which was served in the Thomas
case had been significantly doctored by Ward, and the people in Fujitsu’s litigation
support team had permitted that to happen. We refer to the evidence put to Ward on 1
February 2024 for the details of this shameful episode, and his repeated requests to ask
Jenkins to remove the term ‘system failures’, as it ‘may well support postmasters.”°°
When Jenkins was no longer fit for purpose, Bradshaw, ventriloquised by Cartwright
King’s statement, stepped into the breach. !°°
Over-charging and extracting confessions
138. The practice of charging theft as a threat to bring forth a guilty plea to false accounting
was explained by Paul Whittaker in his statement,!” and exemplified in the case of Janet
Skinner. The truly striking revelation in evidence was that Diane Matthews, the
investigator, did not believe there was any evidence of theft-5* She claims that she raised
this with the lawyer, Juliet McFarlane, but the theft charge was maintained through to a
Plea and Case Management Hearing at the Crown Court, at which point a guilty plea to
false accounting was accepted. As a case study against a backdrop of many other
prosecutions which involved this same “plea bargain” and given that many of the audit
shortfall investigations involved the same essential evidence, there must be many other
cases where the purpose of the theft charge was to act as a threat.
139. There was a slight variation on this theme in the case of Oyeteju Adedayo. The
investigator, Natasha Bernard, did not even believe the “confession” which she had
extracted, and which was used to support the false accounting charge.”
140. Over the course of evidence from many investigators it became obvious that the interview
with the SPM was their focal point, and what they saw as the main part of their job. David
Pardoe testified that there was an expression used to describe those who were not up to
the job of being an investigator: “tackle shy”, meaning that the person was too anxious at
434 FUJ00122197 p6
43° https://www.postofficehorizoninquiry.org.uk/file/2572/download?token=IQIRIA26from p147 onwards
236 INQ00001112 p5 (internal numbering p20)
437 WITN05050100 at §§168-169
238 1NQ00001098 p30 (internal numbering p120)
439 1NQ00001092 p17 (internal number p66)
28
SUBS0000027
the thought of interviewing a suspect. Apparently, such people would be told they were
not right for the role."#°
141. Despite this focus upon interviewing, they were very bad at it. Bernard, for example, had
no knowledge of the “verballing” provisions in PACE, so when it was put to her that she
needed to take care over the signed “confession” that she had been given by the auditor
she was at a loss. Likewise, she had no understanding of s76 of PACE, so could not
recognise the sort of conditions which might make a confession unreliable."
142. There were multiple examples of basing questions on lies, threats and slurs, perhaps the
most extraordinary being the interview of David Blakey by Paul Whitaker, in which
Whitaker suggests that if Blakey does not admit the offending his sick wife will have to
be interviewed, and goes on to suggest that he took the money to spend on a mistress. "7
Tony Utting, Whitaker’s manager, was asked about this interview, and — in essence — he
claims to have seen nothing wrong with it.
143. Aside from this bullying approach to interviewing, there are also many stories of
investigators gratuitously invading the privacy of SPMs and their families, by searching
personal belongings such as underwear drawers and jewellery boxes as if expecting to
find incriminating evidence. Both Vijay Parekh and Vipin Patel have given us instructions
on this and the intimidation felt by their family as a result. There is nothing to suggest
that evidence from searches was ever served in criminal proceedings, which shows that
it was a waste of time, and a practice that would have ended but for the intimidatory
value it offered.
144. Over-charging, intimidation and extracting confessions howsoever possible all drove
towards the same end: if the suspect could be made to provide the evidence against
themselves, there was no need to worry about unreliable IT evidence, and investigators
could send cases to the lawyers without further ado. As the Skinner case shows, it was
sometimes the lawyers who were even more to blame.
Systematising perverting the course of justice
145. It seems that from some time before the introduction of Horizon, Post Office investigators
(then under the auspices of Royal Mail) were preparing two kinds of report: an Offender
Report and a Discipline Report. As described by Tony Marsh, it was always envisaged
that the first of these reports would be sent to the lawyers for advice, and therefore would
attract privilege. The second contained a “subset” of the information in the Offender
Report, and it was envisaged that the Discipline Report would be seen by the defence,
probably in the context of disciplinary proceedings.’ This in itself was a slightly
concerning approach, but at some point, around or before September 2005 the Guide to
completing these reports began saying this:
Details of failures in security, supervision, procedures and product integrity.
This must be a comprehensive list of all failures in security, supervision, procedures and product
integrity it must be highlighted bold in the report. Where the investigator concludes that there
are no failures a statement to this effect should be made ard highlighted in bold.
Significant failures that may affect the successful likelihood of any criminal action and/or
cause significant damage to the business must be confined, solely, to the confidential offender
440 1NQ00001100 p9 (internal numbering 35)
441 1NQ00001092 p16 (internal numbering 61-62)
+2 PQLO0044830 and POLO0044831
143 1NQ00001096 p33-34 (internal numbering p129-133)
+44 1NQ00001068 p26 (internal numbering p103)
29
SUBS0000027
report. Care must be exercised when including failures within the Discipline Report as obviously
this is disclosed to the suspect offender and may have ramifications on both the criminal elements
of the enquiry, as well as being potentially damaging to thereputation or security of the business.
If you are in any doubt as to the appropriateness of inclusion or exclusion you must discuss with
your ITM. (See POLTD. Investigation Circular 2005/1220 Sep 2005)!" [NB we have not been able
to locate this Circular on CP View.]
146. The investigation team was being expressly told not to disclose material which might
undermine prosecutions, and therefore meet the disclosure test, because it might damage
the reputation of the business; and specifically, failures in “product integrity” would
encompass failures in Horizon integrity. Even the choice of word “integrity” suggests
that the authors of this document had Horizon data integrity issues in mind.
147.In 2005 the Designated Prosecution Authority was named in this document as Tony
Utting,"* and the person who drew it up was Ged Harbinson, Compliance Manager.”
The same wording was in the Guide in 29 March 2012, when the Designated Prosecution
Authority was named as Dave Pardoe. However, according to Andrew Hayward, by 2011
all cases were going to John Scott, Head of Security, for a final decision on whether to
proceed.'46
148. The Guide was circulated by David Posnett to a long list of security team members,
alongside the now notorious document giving racist descriptors for ID Codes.'? Neither
Scott nor Posnett had anything to say for themselves when asked about this suite of
documents. Scott pretended that he had no idea they were being circulated, even though
he was perfectly capable of answering coherently when asked specific questions about
the nature of Offender Reports and Discipline Reports; and Posnett made the risible
claim that he had not read the documents he was circulating, and that when completing
the ID Code section of the forms, investigators might use google to find out what the
codes meant, rather than using the Post Office’s own document, which he circulated."
149. This document built upon a policy that was first drafted in March 2000,'* so written just
as Horizon was rolling out. It told investigators to keep information about “security or
operational procedures” confidential in the commercial interests of the business:
Some major procedural weaknesses if they become public knowledge have the potential to assist
others to commit offences against the Post Office, or to undermine the Prosecution case, or to
bring Consignia into disrepute, or to harm relations with major customers such as the DSS or
Girobank. Unless the Offender states that he is aware that accounting weaknesses exist and that
he took advantage of them, it is important not to volunteer that option to the Offender during
interview. The usual duties of disclosure under the Criminal Procedure and Investigations Act
1996 still apply
85 POLOO121569 p11
6 POLO0121569 pS
447 PQLOO121569 p2
448 WITNO8160100 §47
+49 See eg POLOO118104, circulated 31 August 2011
150 1NQ00001083 pp27-9 (internal numbering pp105-116)
451 1NQ00001104 pp39-42 (internal numbering pp115-167)
182 PQLO0104747 pé
453 POLOO104747 p2
30
SUBS0000027
150. This was put to Utting'™ and Posnett'®*. Again, they had nothing to say for themselves.
Utting claimed that the final sentence saved the policy, and Posnett fell back on the usual
lines that he did as he was told, and, in any event, he could not remember anything.
151. We submit that all those involved in the preparation and circulation of these documents
were knowingly engaged in fostering a mindset intent on supressing information which
might undermine the prosecution case, and specifically, information about evidential
problems relating to Post Office IT systems.
PRECIS OF DUNCAN ATKINSON’S KC EVIDENCE
152.5 OCTOBER 2023
© defective practices — mischief of fundamental adverse consequences 49/ 17-24, 51/20-
53/2
¢ lack of training 55/1-25 to 56/1-8
¢ lack of rigour not validating Horizon data 62/7-23
* lack of objectivity as private prosecutor — ‘rubber stamp’ 63/14 to 64/17, 115/18-25
* blurred delineation of functions 89/6-10, 90/3-20, 91/2-6, 162/21 to 163/4
* failure to investigate impartially and secure third-party material 42/12 to 43/11, 46/13
to 47/7, 161/18 to 163/4
* absence of duty of candour 60/10-20
* primacy of business as prosecutor 96/12 to 97/8, 100/19 to 101/line 23
* refusal to countenance system at fault — scapegoating SPMs 178/16 to 179/20
153. 6 OCTOBER 2023
* concerns re expert evidence 59/9-21
* poor CPIA training 85/24 — 87/15, 123/25 to 124/2
* failure to pursue reasonable lines of inquiry 128/15-25 to 129/1-29
© lack of nuance in prosecuting decisions 25/22-25 to 26/1-23
¢ duty of candour not observed 42/6 to 43/18
e inadequate PACE training 121/23 to 123/1
* practice and mischief of charging theft and false accounting 32/20 to 33/21
154. 18 DECEMBER 2023
* prioritising and protecting the business 143/15 to 144/3
* chronic failures re charging and failing to follow reasonable lines of inquiry 15/25 to
17/1, 73/8-15, 81/4-14, 82/9-25, 92/15 to 93/4
* no clear delineation of responsibility 17/8-23, 19/19 to 21/6
* PR elevated to expert evidence 38/14 to 39/8
* approach to confiscation 44/25-46/5
* muzzling and manipulating expert evidence 131/25 to 132/18, 133/35 to 134/15, 135/15
to 136/21, 165/5 to 166/12
155. 19 DECEMBER 2023
« Expert evidence — persistent and extremely serious flaws in approach 35/14 to 36/24,
36/25 to 37/16, 38/5 to 40/13, 133/17 to 134/10, 136/2-23, 165/13-25, 170/7 to 171/5
154 1NQ00001096 p7-8 (internal numbering pp27-31
455 1NQ00001103 p16-7 (internal numbering pp62-5)
31
156.
157.
158.
SUBS0000027
« Absence of cross disclosure — serious concern 4/25-6/8
¢ Failure to pursue reasonable lines of inquiry 131/24 to 132/25
* Inexplicable decision to pursue Seema Misra for theft 13/1-16
e Abuse of Code for Crown Prosecutors 19/16 — 21/23
« Abuse of post-conviction disclosure 23/18 to 24/24, 102/24 to 106/6, 110/24 to 111/5
The findings of Mr Atkinson KC conclusively establish that POL’s disclosure
performance, in policy and practice, was wholly unfit for purpose: “It was looking at how
to protect the system rather than to assess the reliability of the fundamental evidence in
the prosecution of subpostmasters.”1°°
Mr Atkinson expressed “great concern” as to how the approach of POL remained the
same over the period 2002-2012, even though issues regarding Horizon were developing
over that period. The impression, derived from his findings, is that POL battened down
the hatches, or built walls to defend Horizon at all costs. 57 One document is symbolic of
the culture that allowed such lawlessness to flourish.
Likewise, one example of non-disclosure illustrates how POL investigations and
prosecutions worked: when McFarlane became aware that the temporary SPM who took
over after Janet Skinner had been arrested for theft, she recognised that it was disclosable
information, but that information was not in fact disclosed until after the “plea bargain”
had been accepted. This was put to Diane Matthews, who disclaimed responsibility for
this flagrant breach of the CPIA, seeking to blame the lawyer entirely, even though she —
as Disclosure Officer - had not entered the information on the Unused Material
Schedule.” We refer to the refusal by Pardoe, three times, to permit a proper
investigation of the Lyell facts when this was picked up by Mick Matthews at the
confiscation stage’*’ when Janet Skinner had an abuse of process application before the
Crown Court.!*! (This failure to disclose evidence of shortfalls persisting after the
suspension of an SPM is in stark contrast to the service of evidence from temporary SPMs
who claimed not to have experienced shortfalls after the suspension of Castleton and
Misra.)
The R v Seema Misra conspiracy
159.
160.
We submit that there is evidence to support the allegation that Rob Wilson, Jarnail Singh,
Rod Ismay, David Y Smith, John Scott and David Posnett (the Misra group) conspired
together, and with others to pervert the course of justice by prosecuting Seema Misra.
By the time of Seema Misra’s trial in October 2010 the Misra group were deliberately
shutting their eyes to the plain truth that Horizon data was not reliable. They agreed to
pursue the trial without disclosing this truth, and therefore they pursued a course of
conduct which had a tendency to pervert the course of justice, and which was also
intended to pervert the course of justice.
45° In connection with POLO0141396, 18 December 2023 at page 144
157 18 December 2023 at pages 15/25 to 17/1, see also 19/19 to 21/16 as to lack of checks and balances
458 POLO0030622
159 1NQ00001098 p32-34 (internal numbering pp 126-36). Note, too, Natasha Bernard wrongly placing
undermining material on the Sensitive Schedule POLO0026980 in another case in 2009.
16° Henry KC cross-examination of Pardoe INQ00001100 re POLO0064033 at pp.155-162 [pp.39-41/69]
+61 PQLO0048745, POLO0048861. Pardoe forbade Matthews to investigate Lyell on 22 May and 27 Aug.
32
SUBS0000027
161. We reserve the position with respect to Gareth Jenkins’, Andy Dunks and others at
Fujitsu until they have given evidence or chosen not to do so. Likewise Jon Longman.
162.We submit that there are grounds justifying Warwick Tatford’s investigation by his
professional regulator for misconduct.
Stonewalling the defence requests for evidence and disclosure
163. Seema Misra only became aware of the possibility that Horizon had caused the shortfalls
at her branch when Rebecca Thomson’s 2009 article appeared in Computer Weekly. From
that point onwards her lawyers and expert witness made continual, concerted efforts to
obtain full, appropriate disclosure, including four applications to court.'® Nevertheless,
by the time of her trial they did not know about the Known Error Logs, the only bugs
they had been told about were the Callendar Square bug and the ARQ duplication error,
and they had only been provided with ARQ data for a short part of the Indictment period.
164. With respect to Callendar Square, the disclosure was only made because the defence
asked about this bug specifically, given that it had been raised in the Castleton trial. This
can be seen from the Jenkins statement finally served on 1 October 2010.’ The underlying
material was not revealed, not even the original emails which were provided toJenkins.'°°
They showed that, whether or not the bug applied in West Byfleet, the response to the
bug was originally non-existent and later haphazard, at best. The emails admitted that
the bug was in play across the estate for years, affecting many branches on a weekly basis.
This was in stark contrast to the assertion in Jenkins’ statement: “As with any large system
there will be occasional faults such as the one found in Callendar Square, Falkirk. Any
such faults, whether found during testing or from live user feedback would be
investigated and resolved appropriately.”"°°
165. Similarly, the ARQ duplication error was disclosed by way of a sanitising statement from
Jenkins.’ It was presented as if a small, one-off issue had been noticed and readily
corrected. The truth was that this was only one of a series of problems with ARQ data. It
followed swiftly on the heels of the long-running locking problems which are outlined at
§§130-132 above. It was followed by another significant ARQ data integrity issue that
arose in the Derby branch when it migrated to Horizon Online. When that issue was
brought to John Scott’s attention in September 2010, just before the Misra trial, he
effectively closed it down.'® Again, the issue is not just the existence of bugs, but the
handling or lack of handling of them. The defence were not just kept in the dark, but
misled by the assertion that all bugs were “resolved appropriately”.
166. The Misra group did not even provide the basic evidential material underpinning the
Indictment. The Indictment period was 30 June 2005 to 14 January 2008. Originally
Longman requested ARQ data for that period, but he characterised it as a defence request,
182 POLOO175839 reveals that Jenkins (p.2 email to Jarnail Singh, dated 1 March 2010) had gone ‘native’ and
had no notion of being objective or independentwith the first listing for trial then imminent on 15 March 2010
163 YKGI00014994 p3 history set out: first application POLO0093946 heard on 10/3/10 and second
application made 7/5/10; p8 third application made on day 1 of trial ; UKGI00014845 p20 forth application
made on day 6 of trial.
464 PQLO0058440 p21
465 FUJ00083722
46° POLO0058440 pp22-3
187 POLO0061056
468 FYJO0155516
33
SUBS0000027
and it was taken as such by Posnett, who refused it. This was stated to be partly on
grounds of cost, but also because “many cases plead guilty at the eleventh hour and/or
nothing is found by ‘experts’ to challenge the Fujitsu data - the usual attempts of
muddying the waters.”!® Tatford, via Singh, approved the refusal to supply 5 years of
data, on grounds of cost, and said the defence should be required to request and justify
obtaining data for specific periods.” Ultimately, however, at the Inquiry, Tatford
accepted that this was entirely wrong-headed, because the data was prosecution evidence
to support the Indictment they had framed, and it should therefore have been served as
part of their case.17!
167. As the last injury heaped upon insult, there was never any thought given to disclosing
the complete message store. As Chambers said, this was recognised as necessary in the
Castleton litigation, albeit by way of an afterthought, and in a manner which made it
virtually impossible for him to make any use of the disclosure, but it was at least offered.
168. The stonewalling of defence disclosure requests was not even confined to IT. Tatford was
taken to his passage of cross-examination in which he alleged that Mrs Misra was lying
about the trainer who had witnessed her problems with balancing.” She was not lying,
and her team had requested disclosure to support her account,” as it was obviously likely
that the trainer would have recorded what he had witnessed. Nothing was forthcoming,
but the Inquiry has now been given a document which confirms that Michael Opabeyi
approved Mrs Misra putting a sum into local suspense, because he thought a transaction
correction would be forthcoming to account for the inexplicable loss.'4
Abuse of expert evidence
169. In 2010 Tatford was an experienced and competent barrister. He should have been very
well aware of the rules governing expert evidence, and indeed he claims that he was.!°
Nevertheless, under Mr Beer KC’s questioning, Tatford rightly accepted that he had
ignored or broken almost every one of those rules, not least with his interventions on
Jenkins’s Witness Statement; and when it was put to him that he had been trying to
“harden up” the evidence, he accepted that he had “overstepped the mark”.!’° Tatford
repudiated his own Witness Statement for the Inquiry, acknowledging that he had
summarised his interactions with Jenkins in a way which “makes me appear better than
I clearly have been” .!”
170. Grave as it was that Tatford sought to harden up Jenkins’s evidence, he was at least able
to see where he went wrong and make appropriate admissions. When Singh was
questioned on the subject, he also conceded misconduct, but only in an attempt to
maintain the obvious lie that he believed Jenkins to be a witness of fact. When he was
presented with an attendance note in which he recorded Tatford describing Jenkins as
“our expert”, Singh claimed that this was an oversight, and he should have followed up
46° POLO0052202 p2
179 POL00044557 ppS-6
171 1NQ00001094 p8 (internal numbering pp29-30)
172 1NQ00001094 p43 (internal numbering p172)
+73 SMISO000188 p2, POLO0058503
174 POLO0065114
175 1NQ00001094 pp15-6 (internal numbering pp60-61)
176 1NQ00001094 p42 (internal numbering 165 and 166)
177 1NQ00001094 p42 (internal numbering 167)
34
SUBS0000027
on it,!”* but of course the note was only one of many of his own documents which made
it clear that Jenkins was giving expert evidence. His untenable lie was borne of
desperation: how else to explain his real instructions to Jenkins:
As you are our Horizon Expert you need to telephone Charles McLachlan.... to arrange a meeting
where you can discuss all his reports and his concerns about the Horizon so you can deal with it
and rebut it which you have done in your longtelephone conversation about his various
hypothesis and then write a detailed report which would goto someway of progressing and
concluding this matter and importantly preserving the Horizon system.
May be the simplest and practical way of dealing with this whole question is to find a shortest
span of logs, analyse it, disprove or rebut what the Defence Expert is saying in his reports.
Just a reminder you are an Expert for Fujitsu, you will be giving evidence in Court, the Judge
and Jury will be listening to you very carefully and a lot will hang on the evidence.
171. This email shows Singh’s interactions with Jenkins were not the inexperienced,
neutral efforts he sought to suggest they were. It does indicate blundering
incompetence - a more skilful lawyer would have found a more sophisticated way
to manipulate the expert evidence - but it is not in the least neutral. Plainly, by this
point, Singh was engaged in shoring up Horizon whatever the cost, knowing that
he and his colleagues had secured a great many convictions on the back of it. If the
cost was a perversion of the course of justice by deploying a biased Fujitsu “expert”
who would say whatever was necessary to “preserve” Horizon, so be it.
Wilson’s intervention in the Ismay report
172. More troubling still was Wilson’s approach to the question of expert evidence. As Head
of the Criminal Law Team, it was his responsibility to ensure that the lawyers in his
department knew the rules and were capable of instructing experts so that they would be
able to provide independent opinion to assist the administration of justice. When asked
about it, he seemed to suggest that he had never given it a thought:
Q. Did you ever give any guidance of exercise any supervision over the lawyers beneath you in
relation to their professional duties concerning expert evidence?
A. No.
Q. Why was that?
A. I thought we were doing it properly. I don't think I was alive to the problems that you've
pointed out to me.
Q. When you say “alive to the problems’, ie the difference of approach that's needed when you
instruct somebody to give expert evidence as a witness in court proceedings?
A. Exactly.1®
173. The truth is that in March 2010 he had given strategic thought to the use of expert
evidence. At the same time as giving many reasons why he thought it would be a mistake
to instruct an independent expert to review the Horizon system generally, he said this
about the use of expert testimony on a trial by trial basis:
What we really need to do is impress on Fujitsu the importance of fully cooperating in the
provision of technical expertise and witness statements to support the criminal and civil
litigation now and in the future!s!
278 1NQ00001102 p32 (internal numbering p127)
279 POLO0054267
289 1NQ00001106 p14 (internal numbering 55-6)
281 PQ.LO0106867 p1
35
SUBS0000027
174. He knew about the Misra trial, and the unprecedented challenge to Horizon evidence
which it presented, having been party to emails in late 2009 about the disclosure
challenges.!** When he wrote the above sentence a few months later, the Misra case must
have been front and centre in his mind, and he was not much more subtle in expressing
his intent than Singh was when sending his email to Jenkins In trying to avoid the
exposure of wrongful past practice that might come with the introduction of independent
expertise, he evidently thought it would be valuable to obtain “expert” evidence from
Fujitsu, which he rightly expected would shore up the Horizon evidence deployed in
court. If that meant allowing biased and inadequately instructed expertise to pervert the
course of justice in the Misra trial, so be it. This was not only cynical but criminal.
175. We refer to §§63-71 of our Phase 3 Closing Statement for more detail on Wilson and
Ismay’s involvement in the Ismay whitewash and the simultaneous blocking of
disclosure in the Misra proceedings. This was, again we say, not only cynical but criminal.
The Receipts & Payments Mismatch bug
176. This bug came to light shortly before the Misra trial. Jenkins wrote a report on it, dated
29 September 2010, and a document which the Inquiry has now seen many times records
a subsequent meeting about it, at which Jenkins was present.'** These documents
undermined the prosecution case against Mrs Misra, and assisted her defence, partly
because they revealed the presence of a bug which could not be detected by the Horizon
user,’ but also because they revealed a woefully wrong-headed approach to dealing with
bugs. In his report, Jenkins said that if we “amend the data” this will need to be “carefully
communicated” to branches to “avoid questions about the system integrity” !*° The fact
that he expected his readership to accept this speaks volumes. By this point in 2010,
people at working level in POL and Fujitsu expected there to be system integrity issues,
and that SPMs should be kept in the dark about them. The note makes it plain that Fujitsu
and POL expected to be able to resolve this issue covertly.*°
177. These assumptions were further developed in the meeting, and so normalised were they,
whoever took the meeting note felt able to record the three “solutions” under discussion:
SOLUTION ONE - Alter the Horizon Branch figure at the counter to show the discrepancy.
Fujitsu would have to manually write an entry value to the local branch account.
IMPACT - When the branch comes to complete next Trading Period they would have a
discrepancy, which they would have to bring to account.
RISK- This has significant data integrity concerns and could lead to questions of "tampering"
with the branch system and could generate questions around how the discrepancy was caused.
This solution could have moral implications of Post Office changing branch data without
informing the branch.
SOLUTION TWO - P&BA will journal values from the discrepancy account into the Customer
Account and recover/refund via normal processes. This will need to be supported by an
approved POL communication. Unlike the branch "POLSAP" remains in balance albeit wth an
account (discrepancies) that should be cleared.
282 POLO0053723
283 PQLO0028838
284 POL00028838 p7
285 POLO0028838 p8
286 “The Receipts and Payment mismatch will result in an error code being generated which will allow’ Fujitsu
to isolate branches affected this by this problem, although this is not seen by the branches.”POL00028838 p3
36
178.
179.
180.
18
182.
SUBS0000027
IMPACT - Post Office will be required to explain the reason for a debt recovery/ refund even
though there is no discrepancy at the branch.
RISK - Could potentially highlight to branches that Horizon can lose data.
SOLUTION THREE - It is decided not to correct the data in the branches (ie Post Office would
prefer to write off the "lost"
IMPACT - Post office must absorb circa £20K loss
RISK — Huge moral implications to the integrity of the business, as there are agents that were
potentially due a cash gain on their system!*”
The meeting also noted the “Potential impact upon ongoing legal cases where branches
are disputing the integrity of Horizon Data.”'* Sitting in this meeting in early October
2010, Jenkins knew that within days he would be giving evidence in one such legal case.
He must have wondered what the lawyers would expect him to say about this bug and
how it was to be rectified.!*
On the 8 October, which was the Friday before the Misra trial was due to start on Monday
11 October, Alan Simpson in Security emailed the Jenkins report and meeting note to
Wilson, saying “My concern is around the proposed solution/s, one or more of which may
have repercussions in any future prosecution cases and on the integrity of the
Horizon Online system.”'"
Wilson forwarded this email and the attachments to Singh (and McFarlane) at 16.29 that
Friday afternoon. The footer to a print-out of the Jenkins report shows that Singh printed
it from his c: drive at 16.38.'"! Singh knew that Jenkins was due to give evidence at the
Misra trial, and in light of the email from Wilson he must have discussed it with him.
. When asked about this episode, Wilson claimed that he had a meeting with Singh and
McFarlane to discuss the problem of whether and how to tell the SPMs at the affected
branches that their data was to be altered, but not to discuss the much more immediate
problem of whether the bug was to be disclosed in the Misra trial, and what Jenkins was
to say about it.'” It is hard to see why lawyers in the Criminal Law Team would have
anything at all to do with communicating with SPMs who were not involved in any form
of legal proceedings. That was not their responsibility. On the other hand, the decision on
whether to disclose the bug to those SPMs already involved in ongoing proceedings was
very much their responsibility.
In the same passage of evidence Wilson admits that he decided that the Receipts and
Payments Mismatch bug was not disclosable, and accepts that this was wrong, because
the material he received from Simpson should have been disclosed to the Misra defence
(and others being prosecuted). This ties to the inescapable fact that the material received
from Simpson was not disclosed, and Jenkins said nothing about it, neither in discussion
with the defence expert (who would surely have picked up on it if he had), nor in
evidence.
187 POL00028838 p3
488 POLO0028838 p2
189 «
..if he was expressing an opinion that the system worked properly and he was aware of material that
might suggest to the contrary, then he had a duty to disclose that in his report, even if he hadn't been asked
to.” Evidence of Duncan Atkinson KC, 18 December 2023 at pp.133 to 134
190 py
191 py
'OL00055410
'OL00028838 p6
182 1NQ00001106 pp29-30 (internal numbering pp116-120)
37
SUBS0000027
183. Although Singh has not yet answered questions on this subject, in his Witness Statement
he claims not to have seen the Jenkins report on the R&PM bug,'*? and that he knew
nothing of any bugs in the Horizon system.” This despite Wilson saying that the content
of the Simpson email and attachments were “appalling”, which is why he held the
meeting with Singh and McFarlane. We submit there is every reason to conclude that
Singh not only received the Simpson email and printed-out the attachment, but that he
remembers doing so perfectly well, but has lied about it in his Witness Statement.
184. Considering the predicament Singh and Wilson were in on that Friday, there is also every
reason to believe that they decided to tell Jenkins not to say anything about the bug. They
knew he had attended the meeting at which the impact on ongoing legal proceedings was
discussed, and he would therefore be wondering what he ought to say about it. Given
that Wilson admits that he decided the bug was not disclosable, he would not have run
the risk of Jenkins making the disclosure on his own initiative. We submit that Wilson
and Singh conspired to supress the information not only about the bug itself but also
about the implications arising from the Simpson material: 1) that Fujitsu had the power
to alter data covertly in branch accounts, 2) that this meant there had been other bugs
which affected the integrity of Horizon data, and 3) that there were bugs in the system
which were not obvious to Horizon users. As Wilson himself conceded, covert remote
access would have brought all prosecutions to a halt if that had been known. "°°
Absence of Cross-Disclosure: the Hosi case
185. This is another troubling aspect of the matter, especially given the U turn performed by
counsel, who resiled from his initial view and claimed he had been persuaded by Juliet
McFarlane. The chronology appears to be as follows:
e 8 January 2010 - McLachlan requests access to Hosi file and related materials
(FUJ00156097) — Juliet McFarlane asks Jane Owen to find the appropriate person at
Fujitsu to deal with this. At this stage, McFarlane seems far more willing to co-
operate with the defence requests.
¢ There is then a flurry of emails concerned about the implications of the Hosi and
Misra cases (FUJ00152888, FUJ00122701, FUJ00154870)
e 2February 2010 - Juliet McFarlane sends a letter to Warwick Tatford stating “I am a
little concerned at Counsel's suggestion that Royal Mail disclose Hosi's Expert
Report and more particularly the prosecution papers” (POL00053954)
e 4 February 2010 -- Jarnail Singh and Warwick Tatford have a telephone conversation
to discuss disclosure in Misra’s case (FUJ00122794 ~ references this call on page 2)
¢ 5 February 2010 — Jarnail Singh sends an email to the defence stating Counsel has
decided it is not necessary to disclose the preliminary report in the case of Hosi as it
does not pass the disclosure test (POL00054162).
e 5 February 2010 — email from Juliet McFarlane to David Jones (Fujitsu lawyer)
“information regarding the case of Hosi should not be supplied to Mrs Misra's (West
Byfleet) Expert. Indeed Hosi's case should not be discussed with Misra's Expert in
any way without reference to me.” (FUJ00122734)
493 WITNO4750100 §204
194 WITNO4750100 §199
495 Henry KC cross-examination of Wilson 12 December 2023 179/14 to 180/4
38
SUBS0000027
186. It appears from the above that Mr Singh was also implicated in the decision.
The "bandwagon” email and senior reaction to it
187. It is not necessary to quote again from the email which Singh sent at the conclusion of the
Misra trial,'*° but the similarities with Talbot’s crowing email (referred to at §43.c above)
at the conclusion of the Castleton trial are manifest. In both cases the lawyers were keen
to tell those who instructed them within the business that the challenge to Horizon’s
reliability had been vanquished.
188. It is worth noting that there is reason to believe that the content of the email was dictated
to Singh in a phone call: a pro forma front cover sheet which appears to have been fixed
to the Misra file has the text of the email jotted across the corner.” Singh latched on to
this when asked questions about the email,* but gave a typically incoherent and
improbable answer when asked who dictated the email to him, saying that it was a group
of people, possibly including Wilson.!°
189. In any event, the bandwagon email was sent to Rod Ismay, Mike Granville and Mandy
Talbot, amongst others. That ties in to an email Talbot sent to Singh not long before he
received the Simpson email about the R&PM bug, on the Friday before the trial. It told
him that Granville and Ismay were interested in the trial and may give him a call for an
update." No doubt this pressure from above fed into the decision that he and Wilson
made not to disclose the Simpson material.
190. Even more importantly, Ismay passed the bandwagon email on to the relatively new
Managing Director who had commissioned the Ismay Report, David Y Smith. Ismay in
turn passed back this message to the same extensive distribution list that Singh had used
for the bandwagon email:
Dave and the ET have been aware of the significance of these challenges and have been
supportive of the excellent work going on in so many teams to justify the confidence that
we have in Horizon and in our supporting processes. 2°!
Evidently the Executive Team and the Managing Director had been watching the Misra
trial in the aftermath of the Ismay Report. The content of the Ismay Report was so
transparently self-serving, and the Wilson intervention had prevented an independent
expert from feeding into it, so all who had a part in it were implicated in trying to cover-
up the problems with Horizon. It appears from this email that Smith and Ismay were also
191.
actively engaged in subverting the Misra trial as part of the same endeavour. They were
deliberately closing their eyes to problems with the integrity of Horizon data, and were
encouraging their staff to pursue a trial as another method of shoring up the system which
they knew to be deeply problematic.
19° POLOO169170
497 POLO0045121
28 1NQ00001101 p13 (internal numbering p49)
199 1NQ00001101 p14 (internal numbering p54)
2© POLOOOSS418
2°. POL00169170
39
SUBS0000027
Preview of the cover-up conspiracy
192. Phases 3 and 4 have seen glimpses of evidence relating to a cover-up, which started with
the Ismay Report, but gathered momentum in the wake of the Clarke Advice in 2013.
There is much more evidence to be heard on this in Phases 5 and 6, relating to:
a. the Rose Report,
b. the attempted substitution of Bradshaw for Jenkins, despite his manifest inability to
provide expert evidence”,
c. the Horizon meetings, the minuting of them, and the disappearance of Susan
Crichton,
d. the role of Cartwright King and Martin Smith in particular,
e. the suppression of the Horizon spreadsheet of 20 cases by Bolc and others
f. the persistent claim that “remote access” was not possible, the disappearing email
from Lynne Hobbs, and the SLT briefings for Panorama and the Parliament,
g. the commissioning of Detica alongside Second Sight, and the subsequent
suppression of their finding that the Post Office IT systems were “not fit for purpose”.
193. There are also many issues not yet broached: the role of the Government appointed NED
on the Board, the conduct and funding of the GLO, interactions with the CCRC, reviews
by senior lawyers, etc.
194. We submit that this will be evidence of the attempt to cover-up the conspiracies we have
outlined in this Statement. Criminal activity compounded by further crimes against
public justice.
Conclusion
195. When those supposed to uphold the Law, take the Law into their own hands, and break
the Law, in the name of the Law, for their own nefarious purpose, whether for profit, or
brute force of power, there is no Law, only institutionalised lawlessness. This is what
Phase 4 has shown. Those named herein are among those responsible and should be
investigated, fearlessly and independently, in due course. A recommendation from the
Inquiry, to this effect, does not impute liability or guilt, but recognises the weight of
evidence that requires further investigation, and which may require an answer from those
named, in due course.
Edward Henry KC, Mountford Chambers
Flora Page, 23ES Chambers
Hodge Jones and Allen 16 February 2024
202 Contrary to his protestations, denying technical ability, Bradshaw had developed a reputation for vetting
requests for ARQ data, which must have involved him affecting some degree of expertise/overcompetence
regarding that subject. POLO0123286at p.2 reveals that Project Sparrow had exhausted the ARQ quota and so
investigative requests were being rationed, with Bradshaw being praised at being good at dealing withsuch
demands for ARQ data, and seemingly to be put in charge of ‘Special Requests’
40
41
SUBS0000027