Phones 4U Limited (in administration) v EE Limited & Ors

Although there are a number of strands to ground 2 of the appeal, its gravamen is that the judge wrongly proceeded on the basis that, in this case, the Anic presumption could be rebutted by something other than public distancing or a report to the competition authorities. I will consider that point first, after some introductory explanation.

By way of prefatory comment, “public distancing” is a term of art. As the judge explained at [220], rather than requiring something to be done publicly in the normal sense, it means that the undertaking makes clear to the other parties involved that it wishes to take no part in the actual or proposed anticompetitive conduct. For convenience, I will also use the term to encompass the alternative of a report to the competition authorities.

The Anic presumption was established in Case C-49/92P Commission v Anic [2001] 4 CMLR 17 at [121]:

“… subject to proof to the contrary, which it is for the economic operators concerned to adduce, there must be a presumption that the undertakings participating in concerting arrangements and remaining active on the market take account of the information exchanged with their competitors when determining their conduct on that market…”

It is important to distinguish this from a different presumption, also discussed in Anic but considered in more detail in Case C-204/00 P Aalborg Portland A/S v Commission [2005] 4 CMLR 4 (“Aalborg Portland”). In Anic at [96], the Court had held that, since the Commission had been able to show Anic’s participation in price-fixing meetings, “it was for Anic to adduce evidence that it had not subscribed to those initiatives”. In Aalborg Portland, the principle was explained as follows:

“81.

According to settled case law, it is sufficient for the Commission to show that the undertaking concerned participated in meetings at which anti-competitive agreements were concluded, without manifestly opposing them, to prove to the requisite standard that the undertaking participated in the cartel. Where participation in such meetings has been established, it is for that undertaking to put forward evidence to establish that its participation in those meetings was without any anti-competitive intention by demonstrating that it had indicated to its competitors that it was participating in those meetings in a spirit that was different from theirs.

82.

The reason underlying that principle of law is that, having participated in the meeting without publicly distancing itself from what was discussed, the undertaking has given the other participants to believe that it subscribed to what was decided there and would comply with it.

83.

The principles established in the case law cited at [81] also apply to participation in the implementation of a single agreement. In order to establish that an undertaking has participated in such an agreement, the Commission must show that the undertaking intended to contribute by its own conduct to the common objectives pursued by all the participants and that it was aware of the actual conduct planned or put into effect by other undertakings in pursuit of the same objectives or that it could reasonably have foreseen it and that it was prepared to take the risk.

84.

In that regard, a party which tacitly approves of an unlawful initiative, without publicly distancing itself from its content or reporting it to the administrative authorities, effectively encourages the continuation of the infringement and compromises its discovery. That complicity constitutes a passive mode of participation in the infringement which is therefore capable of rendering the undertaking liable in the context of a single agreement.”

Therefore, participation in a meeting at which an anticompetitive agreement is concluded without public distancing is presumed to amount to participation in it, because the undertaking in question is treated as providing tacit approval. The same presumption would apply in relation to the concertation element of a concerted practice, as the Eturas decision discussed below illustrates. To distinguish this presumption from the Anic presumption I will refer to it for convenience as the “participation presumption”.

The foundation for Mr MacLean’s submission that public distancing was required to rebut the Anic presumption is the CJEU’s judgment in Eturas at [46]. That is said to be authority for the proposition that, where the alleged collusion occurs at a meeting (such as, in this case, the Landmark lunch), the Anic presumption can only be rebutted by public distancing. Mr MacLean also relies on what he says was an endorsement of this by the Supreme Court in Sainsbury’s Supermarkets Ltd v Mastercard Inc; Asda Stores Ltd v Mastercard Inc; Sainsbury’s Supermarkets Ltd v Visa Europe Services LLC [2020] UKSC 24, [2020] 4 All ER 807 (“Sainsbury’s v Visa”) at [113].

I do not agree. On a proper reading, Eturas does not provide support for the proposition, and P4u has not identified any CJEU authority that does. The proposition lacks logic and would lead to unprincipled differences of approach depending on whether there had or had not been a meeting. Further, Sainsbury’s v Visa concerned a very different issue.

Eturas concerned an online booking system called E-TURAS which was used by travel agents in Lithuania. Participating firms were sent an electronic message informing them that the maximum discount that could be applied for bookings made using the system was being reduced to 3%. The modified system made it harder to grant a higher discount, though it was still technically possible. The national competition authority concluded that participants who continued to use the system without objection following receipt of the message had participated in an anticompetitive concerted practice.

The CJEU reformulated the questions referred by the Lithuanian court as whether “it may be presumed that [the travel agents] were aware or ought to have been aware of that message and, in the absence of any opposition on their part to such a practice, it may be considered that those operators participated in a concerted practice” within Article 101(1) (Eturas at [26]). The answer was that the firms “may – if they were aware of that message – be presumed to have participated in a concerted practice…, unless they publicly distanced themselves from that practice, reported it to the administrative authorities or adduce other evidence to rebut that presumption, such as evidence of the systematic application of a discount exceeding the cap in question” (see at [50] and the dispositif).

It is notable that in Eturas a concerted practice obviously existed as between the operator of the system and those who were aware of the new arrangement and went along with it. The question was whether firms should be treated as having participated in that concerted practice even if they were not aware of the message. This has nothing to do with the Anic presumption. Rather, it relates to the participation presumption referred to in Anic at [96] and discussed further in Aalborg Portland: see above. Further and in any event, the CJEU was obviously not concerned with meetings.

Reflecting the issue in the case, the CJEU’s discussion focuses on the question of participation in what was undoubtedly a concerted practice for those who knowingly went along with it. At [28] the CJEU explained that “passive modes of participation”, such as being present in meetings at which anticompetitive agreements are reached, are indicative of collusion, since (absent public distancing) the tacit approval it provides encourages the infringement to continue. That reflects Aalborg Portland at [84]. The CJEU then went on at [29] to [49] to discuss three considerations in that regard, the first being burden of proof, the second being the three elements required for a concerted practice (see [18] above) and the third being the ability to rebut the presumption of participation. A careful reading of these paragraphs does not support Mr MacLean’s submissions.

In relation to the first consideration, burden of proof, the CJEU drew a distinction at [34] between the Anic presumption (summarised at [33]), which formed an integral part of EU law, and the question of awareness, which was one for national law but subject to EU law considerations, including the potential for a concerted practice to be inferred otherwise than from direct evidence and the presumption of innocence. As to the second, the CJEU set out the three elements of a concerted practice at [42] and then referred to the facts of the case before observing at [44] that they were capable of justifying a finding of concertation between firms aware of the message, provided the other two elements were satisfied, and that depending on the evidence “a travel agency may be presumed to have participated in that concertation if it was aware of the content of that message”. In contrast, if awareness was not established then, unless there was other evidence showing tacit assent, the firm’s participation could not be inferred ([45]).

The CJEU then said this as to the third consideration:

“46.

In the third place, it must be pointed out that a travel agency may rebut the presumption that it participated in a concerted practice by proving that it publicly distanced itself from that practice or reported it to the administrative authorities. In addition, according to the case-law of the Court, in a case such as that at issue in the main proceedings, which does not concern an anticompetitive meeting, public distancing or reporting to the administrative authorities are not the only means of rebutting the presumption that a company has participated in an infringement; other evidence may also be adduced with a view to rebutting that presumption (see, to that effect, judgment in Total Marketing Services v Commission, C‑634/13 P, EU:C:2015:614, paragraphs 23 and 24).

47.

As regards the examination of whether the travel agencies concerned publicly distanced themselves from the concertation at issue in the main proceedings, it must be noted that, in particular circumstances such as those at issue in the main proceedings, it cannot be required that the declaration by a travel agency of its intention to distance itself be made to all of the competitors which were the addressees of the message at issue in the main proceedings, since that agency is not in fact in a position to know who those addressees are.

48.

In that situation, the referring court may accept that a clear and express objection sent to the administrator of the E-TURAS system is capable of rebutting that presumption.

49.

As regards the possibility of rebutting the presumption of participation in a concerted practice by means other than public distancing or reporting it to the administrative authorities, it must be held that, in circumstances such as those at issue in the main proceedings, the presumption of a causal connection between the concertation and the market conduct of the undertakings participating in the practice, referred to in paragraph 33 of the present judgment, could be rebutted by evidence of a systematic application of a discount exceeding the cap in question.” (Emphasis supplied.)

Mr MacLean relied on the words underlined in [46] to support P4u’s argument that, where there has been a meeting, the only means of rebutting the Anic presumption is public distancing. However, those words do not relate to the Anic presumption at all, but instead to the participation presumption. This is apparent from the context, including the question that the CJEU was addressing in that case and the separate specific reference to the Anic presumption at [49] (in addition to the reference to the participation presumption at the start of that paragraph), and also from [46] itself. That paragraph not only refers to a presumption of participation (“the presumption that a company has participated in an infringement”), but cites as authority Total Marketing at [23] and [24].

In Total Marketing, the CJEU explained at [20] that public distancing was required to show that participation in collusive meetings was without anticompetitive intent (citing Aalborg Portland at [81] and [82]), that there was a presumption of illegality of such participation which must be rebutted through public distancing ([21]), but that it followed that public distancing was only necessary to rebut the presumption in the case of an undertaking that participated in anticompetitive meetings ([22]). In other circumstances the absence of public distancing forms “only one factor amongst others to take into consideration”, and the General Court had therefore erred in concluding that public distancing was “the only means available to an undertaking involved in a cartel of proving that it has ceased participating in that cartel, even in the case where that company has not participated in anticompetitive meetings” ([23] and [24]). None of this related to the Anic presumption.

It makes logical sense that presence at a meeting at which collusion occurs should be regarded as participation in that collusion unless public distancing occurs. The rationale for that is explained in Aalborg Portland (see above). The same logic does not necessarily apply outside the context of a meeting.

In contrast, there is no logic in applying an equivalent approach to the Anic presumption. Whether collusion occurred at a meeting or in some other way should make no difference to the question of cause and effect between concertation and subsequent conduct on the market. Similarly, there is no principled reason why it should be impossible to rebut the Anic presumption by anything other than public distancing where there has been a meeting, however compelling the evidence is that there was in fact no causal effect. It would also lead to bizarre distinctions between meetings (presumably in person or virtual) and, for example, WhatsApp chat messages.

As Mr Pickford pointed out, P4u’s interpretation of Eturas at [46] is also not supported by other CJEU decisions. The formulation of the test in Case C-455/11P Solvay SA v Commission [2014] 4 CMLR 17 (“Solvay”) at [43], set out at [171] below, makes no mention of it. Further, the facts of that case involved meetings, yet the CJEU considered whether other evidence could rebut the Anic presumption. Similarly, Case C-8/08 T-Mobile Netherlands BV v Raad van bestuur van de Nederlandse Mededingingsautoriteit [2010] Bus LR 158 concerned a meeting, but the CJEU’s reference at [61] to evidence required to rebut the presumption is in general terms. Case C-251/22P Scania AB v Commission [2024] 4 CMLR 24 does not take the matter any further in its references to both the participation presumption and the Anic presumption at [99]-[103].

Sainsbury’s v Visa does not compel a different conclusion. The relevant issue there related to the standard of proof and the nature of evidence required to meet the exemption in Article 101(3). In outline, the Supreme Court held that the former was a question of domestic law but the latter was subject to EU law requirements.

In Sainsbury’s v Visa, both Mastercard and Visa relied on Eturas at [30]-[32] to argue that the Court of Appeal had applied an unduly onerous standard of proof. The Supreme Court observed at [113] that this passage was not as helpful to Mastercard and Visa as might first appear because it concerned proof of awareness of the content of the message, which was unsurprisingly a matter for national law, but it added:

“The Court of Justice went on, however, (at para 33) to distinguish the presumption arising under art 101(1) of a causal connection between a concertation and the market conduct of the undertakings participating in the practice. That presumption, it emphasised, followed from art 101(1) and consequently formed an integral part of the EU law which the national court was required to apply. The Court of Justice then went on (at paras 46-49) to address in detail the nature of the evidence that would be sufficient to rebut the presumption.”

This obviously refers to the Anic presumption. However, the cross-reference to Eturas at [46]-[49] is not entirely consistent with the analysis above. On that analysis, [46] refers to the participation presumption and [49] refers both to that presumption and the Anic presumption.

Ultimately I have concluded, with respect, that there is a minor slip in the Supreme Court’s judgment. There was no need in that case to consider the two different presumptions at play. That there was some confusion between them is supported by what the Court went on to say at [114]:

“In our view, the fact that the Court of Justice in Eturas addressed, as a question of EU law, what evidence was capable of rebutting the presumption of participation in a concerted practice provides the key to resolving the present issue.”

The Court went on to refer to paragraph 100 of the Advocate General’s opinion in Eturas, part of a section which discusses rebuttable presumptions in competition law generally.

P4u also makes other criticisms of the judge’s approach to the Anic presumption. It maintains that he wrongly held that the presumption applied only to those in receipt of information (EE in the case of the Landmark lunch) and that he failed to apply the demanding case law test for rebuttal. I will deal with the second of these first.

Mr MacLean submitted that, even if EE was entitled to rely on evidence other than public distancing to rebut the Anic presumption, the judge failed to apply the exacting standard in Solvay at [43], namely that:

“…it is for the undertaking concerned to prove that the concerted action did not have any influence whatsoever on its own conduct on the market… The proof to the contrary must therefore be such as to rule out any link between the concerted action and the determination, by that undertaking, of its conduct on the market.”

Solvay was not cited at the trial.

P4u’s case is that, if this high threshold was applied, then EE could not meet it. The judge had focused on the immediate aftermath of the Landmark lunch, rather than EE’s ultimate decision in 2014 to exit P4u. It was impossible to rule out an indirect link between the concertation and that later decision. This was because EE would have had the benefit of knowing that, having regard to its anti-competitive approach, O2 must have been so committed to strategic action against indirect retailers that it could probably be relied upon not to recommence supplies at a later date.

The judge was entirely right to give this short shrift in the permission judgment.

First, the standard of proof remains the usual domestic law one, namely the balance of probabilities: Sainsbury’s v Visa at [115] and [116].

Secondly, the factual allegation that EE gained comfort that O2 would not re-enter the market was not advanced at the trial and was not put to Mr Swantee. It should have been if P4u wished to rely on it.

Thirdly, the judge was in any event correct to reject it. When EE made its own decision to leave P4u in mid-2014 it was of course aware that by then O2 had ceased to supply P4u both as regards new connections (from 31 January 2013) and upgrades and SIM-only connections (from 31 January 2014). That was public knowledge. In contrast, at the time of the Landmark lunch there was only a provisional decision to leave P4u but, more importantly for present purposes, the judge found that what Mr Dunne conveyed was limited and vague. When he learnt later of O2’s actual departure from P4u, Mr Swantee may well have been able to put two and two together in relation to that, but I do not consider it realistic to suggest that, when EE came to make its own decision in 2014, what Mr Dunne had said almost two years earlier – and obviously in different market circumstances – could provide any comfort as to what O2 might do thereafter, beyond what could be gleaned from the concrete fact of what was by then its complete exit from P4u.

Mr MacLean also submitted that the judge erred at [232] by failing to recognise that the Anic presumption applies to all participants in concertation, and not just to recipients of information.

This is a mischaracterisation of what the judge said at [232], where he referred to the presumption not applying to the party disclosing the information and said that, on the basis of his finding that “Mr Swantee made no disclosure to Mr Dunne of EE’s position, nor did he offer to give any support from EE for O2’s strategy” there was no scope for the presumption to apply as regards O2 ([41] above). The critical point was that Mr Swantee conveyed nothing to Mr Dunne, so that there was nothing that Mr Dunne gleaned from the meeting which could have influenced O2’s conduct on the market: see ground 1, above. In contrast, if Mr Swantee had given some indication or comfort that he was prepared to cooperate or support the move, then that would in effect have been a provision of information by Mr Swantee to Mr Dunne, with scope for the Anic presumption to apply as regards O2’s conduct thereafter: see the discussion at [146] above.

Finally under this ground, P4u alleges that the judge erred at [234] in concluding that, even if there had been an infringement, it had no causal effect on O2’s decision to leave P4u because he wrongly put the Anic presumption to one side at that stage.

Given that the judge made no error in concluding that there was no infringement it is not necessary to address this point. I will just note that it faces the hurdle of the principle that causation and loss are matters for domestic law: Deutsche Bahn AG v Morgan Advanced Materials plc [2014] UKSC 24, [2014] 2 All ER 785 at [11].

It follows that I would dismiss ground 2. As a result, the judge’s conclusions that the Anic presumption was rebutted as far as EE was concerned in relation to the Landmark lunch and by Vodafone in relation to the meeting between Ms Castillo and Mr Humm (see [41] and [49] above, and also [269] below) both stand.

Ground 3 relates to the alleged collusion between EE and Vodafone in 2014 and the alternative explanation posited by the judge as to the source of information about Vodafone being “about to pull out”, referred to in Mrs Derbyshire’s email of 10 August 2014 (see [92]-[96] above). P4u maintains that the judge impermissibly rejected P4u’s case based on a factual theory that had not been canvassed at trial. Rather, following the principles set out in Al-Medenni v Mars UK Ltd [2005] EWCA Civ 1041 (“Al-Medenni”) it was not open to the judge to advance this new theory. P4u relied on the 10 August email as evidence that there had been earlier collusion, in particular between Mr Scheen and Mr Humm in their call on 2 April 2014.

Al-Medenni concerned a claim against Mars for injuries suffered by an employee when a roll of wrapping paper fell from a machine. Her claim pleaded that it had been placed there by a specific individual, a Mr Braich, whereas Mars defended the claim on the basis that the claimant had put it there herself. During opening submissions the trial judge raised the possibility that a third person had placed the roll on the machine. The point was not put to witnesses, but following a hint from the judge counsel for Mrs Al-Medenni adopted this unpleaded “third man theory” as an alternative in her closing submissions. The judge found for Mrs Al-Medenni on the basis of the third man theory, finding that an unidentified employee was responsible.

The Court of Appeal allowed Mars’s appeal. Dyson LJ, with whom Brooke and Tuckey LJJ agreed, said at [21]:

“In my view the judge was not entitled to find for the claimant on the basis of the third man theory. It is fundamental to our adversarial system of justice that the parties should clearly identify the issues that arise in the litigation, so that each has the opportunity of responding to the points made by the other. The function of the judge is to adjudicate on those issues alone. The parties may have their own reasons for limiting the issues or presenting them in a certain way. The judge can invite, and even encourage, the parties to recast or modify the issues. But if they refuse to do so, the judge must respect that decision. One consequence of this may be that the judge is compelled to reject a claim on the basis on which it is advanced, although he or she is of the opinion that it would have succeeded if it had been advanced on a different basis. Such an outcome may be unattractive, but any other approach leads to uncertainty and potentially real unfairness.”

Dyson LJ went on at [22] to explain that the “starting point” was the pleadings. He referred to comments by Lord Phillips MR in Loveridge and Loveridge v Healey [2004] EWCA Civ 173 at [23] to the effect that pleadings mark the parameters of the case and form the basis of decisions about preparations for trial, including about the evidence required, and that where:

“…departure from a pleading will cause prejudice, it is in the interests of justice that the other party should be entitled to insist that this is not permitted unless the pleading is appropriately amended. That then introduces, in its proper context, the issue of whether or not the party in question should be permitted to advance a case which has not hitherto been pleaded…”

In Al-Medenni there had been no application to amend and the “rather faint-hearted” espousal of the theory in closing submissions was far too late ([23]). The defendants were denied the opportunity to explore the theory with the witnesses or to consider seeking evidence from others ([24]). Rather, the judge should have concluded that the claim was not proved. The judge had “crossed the line which separates adversarial and inquisitorial systems” ([25]).

Satyam Enterprises Ltd v Burton [2021] EWCA Civ 287, [2021] BCC 640 (“Satyam”) concerned the reverse situation of an unpleaded defence being adopted by the trial judge. There the claim advanced was that a former director and sole shareholder of the claimant company had disposed of its assets at an undervalue. The defendant defended the claim on the basis that there was no undervalue. He also pleaded that he had held the company’s shares on trust for a third party, a Mr Sharma. The judge dismissed the claim on the basis that the properties disposed of by the company were themselves held on trust for Mr Sharma, a proposition which had formed no part of either party’s case and was not raised at trial.

Nugee LJ, with whom Lewison and Arnold LJJ agreed, said at [36]:

“The present case … is not one of a party seeking to depart from his pleaded case, but one where the parties addressed in their evidence and submissions the cases that had been pleaded, but the judge decided the case on a basis that had neither been pleaded nor canvassed before him. In our system of civil litigation that is impermissible, and a misunderstanding of the judge’s function which is to try the issues the parties have raised before him…”

Nugee LJ then considered Al-Medenni before concluding that the course taken by the judge was not open to him, adding at [38]:

“Judges may sometimes think—and may even sometimes be right—that their own theory better fits the facts than that of either party, but if it is wholly outside the scope of the pleaded issues, that is nothing to the point, and to decide a case on a basis that has not been explored in evidence or addressed in submissions is likely to leave at least one, if not both, parties with a profound and justified sense of unfairness.”

Al-Medenni and Satyam were considered again in Ali v Dinc [2022] EWCA Civ 34. In that case the trial judge had reached a conclusion based on an “intermediate” combination of factual assertions from each party’s case, a combination which neither had expressly pleaded. The defendants unsuccessfully appealed.

Birss LJ, with whom Green and Whipple LJJ agreed, added the following at [25] to what had been said in Al-Medenni and Satyam:

“These problems are all concerned with the interests of justice and, in particular, with circumstances which cause prejudice to the losing party. The common sort of prejudice which is to be avoided is that a new point has arisen in such a way that the losing party was not given a proper chance to call evidence or ask questions which could have addressed it. That is why the function performed by pleadings, lists of issues and so on, which is to give notice of and define the issues, is an important one; but is also why a judge can always permit a departure from a formally defined case where it is just to do so. It is also why the judge’s function is to try the issues the parties have raised before them, rather than to reach a conclusion on the basis of a theory which never formed part of either party’s case. By placing the emphasis on prejudice, the point I am making is that the modern approach to the definition of the issues requires judges to adopt a pragmatic approach in line with the overriding objective and not seek to be governed by unnecessary formality, provided always that it is just not to do so.”

In Ali v Dinc, the judge reached a decision on the pleadings that was open to her. Her conclusions were “composed entirely of the acceptance or rejection of factual assertions which were pleaded” ([29] and [30]). Further, in rejecting an argument from the appellants that the claimant had narrowed the judge’s options by abandoning an alternative case, Birss LJ also observed that it could not be said that the appellants were ambushed or precluded from advancing submissions or evidence that they might otherwise have done: no prejudice had been identified ([34]).

Although the focus in Al-Medenni and Satyam was on the pleadings, Birss LJ’s comments make clear that the key determining factor is the interests of justice, and specifically the question of prejudice. So, for example, it could be that an uncanvassed theory advanced by a judge in relation to an important piece of evidence could be regarded as sufficiently unfair to warrant intervention by an appellate court, even though (being evidence rather than the facts necessary to establish a claim or defence) it did not form, or need not have formed, part of the pleadings. Conversely, I would not rule out the possibility of some departure from pleaded issues as not giving rise to unfairness on the particular facts of a case, although I would expect that to be rare unless the position was fully addressed at trial.

Drawing the threads together, I would summarise the position as follows:

It follows that I would not accept Mr Pickford’s submission of a “bright line” distinction between a departure from the pleaded issues and other evidence, the former being always impermissible. I would however agree that there is a spectrum, with cases of a wholesale departure from a pleaded case, as in Al-Medenni and Satyam, being at an extreme end of decision-making that will clearly be impermissible.

An analogy can be drawn with the approach that appellate courts take to raising new points on appeal, summarised by Lewison LJ in Hudson v Hathway [2022] EWCA Civ 1648, [2023] KB 345 at [34] and [35]. An appellate court will not generally permit a new point to be raised if it would necessitate new evidence or would have resulted in the trial being conducted differently with regards to the evidence. Further, even a new point of law may only be raised if (among other things) the other party has had adequate time to deal with it. However, there is a spectrum of cases and the just outcome will depend on an analysis of all relevant factors, including the question of prejudice.

As a general rule, judges should be very cautious about advancing a theory which has not been put forward by the parties. If it is outside the scope of the pleaded cases it should generally not be done at all: see above. If, short of that, a judge concludes that it is necessary to his decision to address such a theory in relation to any material evidence, the proper course is to raise it with the parties and to provide them with an adequate opportunity to make submissions about it, including as to whether further evidence is required to ensure a just outcome.

Although it was in the category of evidence rather than facts that needed to be pleaded to establish a cause of action (which is the proper scope of particulars of claim), P4u did rely on Mrs Derbyshire’s email dated 10 August 2014 in its re-amended particulars of claim. Specifically, it relied on the email (and a comment in the presentation for the 21 May 2014 BRM) to support an allegation of communications between EE’s parents, DT and/or Orange, and Vodafone, which was then “reflected in EE’s expectation” that Vodafone UK would cease to supply P4u. In response, EE pleaded that Vodafone UK’s future conduct was unknown to it and denied that it had any generally held expectation that Vodafone UK was about to discontinue trading with P4u, but also pleaded that “EE did not have the alleged knowledge and/or understanding obtained from Vodafone UK and/or Vodafone Group”. For its part, Vodafone denied that any view EE had formed resulted from coordination or collusion with it.

The judge found that confidential information had been received, but not from Vodafone. In effect, the judge rejected EE’s main defence (that no such information had been received) but accepted both Vodafone’s defence and EE’s pleading that Vodafone had not been the source of any such information. This was therefore far from the facts of Al-Medenni or Satyam. This was a question of evidence rather than a fact that was necessary to establish a cause of action, and moreover the judge did not go outside the ambit of the pleadings in deciding the point.

Nevertheless, the question remains as to whether there was prejudice to P4u such as to warrant interference in the judge’s decision. Further, it is necessary to consider that question in the context of the other grounds of appeal, and in particular grounds 4, 5 and 7, because the assessment of prejudice in this case depends in part on the judge’s other factual findings. My conclusions on this ground accordingly depend in part on the view I have reached about those other grounds, discussed below. As noted at [227] below, the same applies in reverse.

The arguments for P4u on this point appeared at some points to suggest that, having concluded that EE did receive confidential information, the judge was bound to find that it came from Vodafone (contrary to Vodafone’s pleaded case), because no alternative had been advanced by the parties. That cannot be right. A judge cannot be required to conclude (positively) in favour of a factual case that one party is putting forward even if the judge disagrees with it because it does not accord with his own interpretation of the evidence, including as to the inherent probabilities. That is not what either Al-Medenni or Satyam is authority for.

Further, the judge correctly pointed out at [605] that before drawing inferences the court should consider surrounding circumstances to see whether there is another plausible explanation. This was a point that the judge had first made earlier in the judgment in the context of a self-direction about the sparse evidence usually available about unlawful contacts between traders and the need to rely on inferences to determine their existence ([84]-[86]). He cited there a statement in Eturas at [36] which refers to “a number of coincidences and indicia which, taken together, may, in the absence of another plausible explanation, constitute evidence of an infringement of the competition rules”, and added at [86] that:

“…before drawing inferences the Court must be careful to ensure that there is no equally plausible and innocent explanation for the fragmentary evidence on which reliance is being placed. To do otherwise would be to reverse the burden of proof as regards serious allegations. Altogether, I consider that the Court has to consider the evidence in the round, looking at the particular items of evidence relied on in context.”

I agree. It is true that, for obvious reasons, there may be a lack of direct evidence of unlawful activities and therefore a need to rely on drawing inferences from circumstantial evidence (see for example Aalborg Portland at [56]-[57], reflected in Eturas). However, the recognition of that point does not remove the need for the allegations to be proved on the balance of probabilities.

As already noted, the question is whether there was prejudice to P4u such as to warrant interference in the judge’s decision. For the reasons that follow I have concluded that there has been no such prejudice. However, that does not mean that the judge took an appropriate course. Assuming the possibility that CPW was the source of the information occurred to the judge only after the trial and he considered it sufficiently important to refer to it, the better approach – however unattractive in practice – would have been to raise it with the parties and provide them with the opportunity of making submissions about it, including as to whether any further evidence was required to ensure a just outcome.

Further, I do not consider that the respondents are materially assisted by the judge’s emphasis at [605] that he was not making a positive finding that CPW was the source of the information. The existence of that possibility, which the judge found to be “more plausible” than P4u’s case as to the source of the information, appears to have been a key reason to reject the contention that Vodafone was the source.

Mr MacLean submitted that, if the possibility of CPW being a source had been “in play”, it could have sought disclosure of relevant communications and considered whether to seek witness evidence from CPW employees, as well as otherwise exploring the point in the evidence. Further, the judge’s alternative reasoning, including that there was no causal effect (see [95] and [96] above) was based on EE obtaining information in August 2014, whereas P4u’s case was that the 10 August email evidenced collusion at an earlier date, in April.

Mr Pickford submitted that P4u always needed to show both that the email reflected confidential information rather than speculation and that the information came from Vodafone. On the second point, it should have been obvious that it would assist P4u if it could discount other sources. Indeed, P4u’s pre-action correspondence showed that its solicitors had considered CPW as a source and indeed had raised that with CPW’s solicitors. It was unsurprising that that was not pursued because it would have harmed P4u’s case to consider CPW as a source. Mr Williams also emphasised Vodafone’s pleaded defence that it was not the source of the information, and that P4u had every chance to prove its case to the contrary but failed to do so.

There is strength in each of these submissions, but they are not determinative.

One feature of the judge’s decision on this point is that there would have been no legitimate complaint if the judge had limited his reasoning to concluding, based on the witness evidence and inherent probabilities, that Vodafone was not the source of the information rather than speculating about other possible sources. The difficulty has arisen because he went further. However, on analysis he did so in a way that was immaterial to his decision that Vodafone and EE did not collude in April 2014.

My reading of the judge’s findings at [603]-[607] is that he was focusing on the email as reflecting a provision of information in early August. I agree with Mr Williams’ submission that that was the context in which the judge concluded that there was a more plausible explanation, and then made further comments about later modelling and Mr Eyre’s evidence about what he would have done in August if only he had known what Vodafone was about to do ([95] above). Thus, the theory that CPW was the source assumes, contrary to P4u’s case, that the information was provided to EE in August, not April.

It is only at [608] that the judge refers to P4u relying on the email to support the allegation of earlier collusion. That allegation had been rejected in the judge’s earlier, lengthy, analysis of Vodafone’s and EE’s respective decision-making processes, but was not explicitly re-addressed in the context of the email from Mrs Derbyshire. The reason for this is apparent from [608] of the judgment, including its reference to “insofar as EE gained information in early August as to Vodafone UK’s intention as regards P4u”. In other words, and consistent with his reasoning at [603]-[607], the judge did not read the email as referring to anything other than recently obtained information. I do not need to rely on it, but for what it is worth I note that the permission judgment also supports the fact that the judge considered that the information was recently obtained (see [36] and [37] of that judgment).

I do not find it at all surprising that the judge’s focus was on the position in early August. It is inherently unlikely that a statement that the “expectation is that Voda are about to pull out” could relate to information provided months earlier, as opposed to recently acquired information. Further and importantly, as Mr Pickford pointed out, the extract from the email that is relied on refers to “the impact if P4u fold earlier than expected ie In the next couple of months” (emphasis supplied). EE’s expectation in May 2014 had been that P4u would remain in business at least until 2016: see [77] and [78] above, discussed further below in relation to ground 4. This is clearly supportive of the information having been received more recently than then.

The judge had dealt in exhaustive detail with the developing positions of EE and Vodafone during the earlier period, as recorded in contemporaneous documents, and concluded that no collusion had occurred: see the summary of the judgment, above. The effect of his findings was that EE made no decision to leave P4u before 21 May 2014 at the earliest, and Vodafone only on 17 June 2014 (see [61] and [79] above). Properly understood, the 10 August email was in contrast a very slender piece of evidence to counter the evidence relied on by the judge in reaching those other conclusions. In particular, the suggestion that information was obtained in early April indicating that “Voda are about to pull out” is very hard to fit with the process by which Vodafone reached its decision in June, a decision which was only made after achieving successful engagement with CPW from April onwards to agree the terms of an exclusive deal, as well as obtaining approval from Vodafone Group ([58]-[61] above). In contrast, what Mrs Derbyshire said fits well with the information being received in early August.

Given P4u’s reliance on it, it would have been preferable if the judge had addressed Mrs Derbyshire’s email more explicitly in the context of the collusion alleged to have occurred between EE and Vodafone in April 2014, by spelling out why he considered that the email should be read as reflecting only recently obtained information. However, the reality is that the new case theory advanced by the judge was not relied on by him to reject the allegation of collusion on which P4u actually relied, but to explain where he thought the information could have come from in early August. The allegation on which P4u relied was rejected earlier in the judgment based on the judge’s analysis of other evidence. He was entitled to do so, and nothing in the email casts doubt on his conclusion.

I would therefore dismiss ground 3.

Grounds 4 and 5 both relate to areas of fact-finding where it is said that the judge erred. P4u relies on the delay in handing down the judgment to justify the application of a lower hurdle than is usually applied to challenges on appeal to findings of fact. The challenge under ground 4 is that the judge failed to consider or take into account, or mis-recollected, material evidence. Ground 5 is that the judge adopted a “compartmentalised” approach, failing to consider the evidence in the round. Some aspects of the complaints under ground 4 overlap heavily with ground 5, so I will deal with the two grounds together. In summary, I have concluded that each of the challenges should be dismissed.

The effect of a delay in producing a judgment has been considered by this court on a number of occasions, most recently in NatWest v Bilta and a year earlier in Bank St Petersburg PJSC v Arkhangelsky [2020] EWCA Civ 408, [2020] 4 WLR 55 (“Bank St Petersburg”). Both of those cases also considered challenges based on “compartmentalisation”. Rather than repeating all of what is said there, I will summarise some key points and make some additional observations.

As to delay:

I would add the following observations to these points.

First, one of the points made by Lewison LJ in Fage v Chobani at [114] was the fact that the trial judge will have had regard to the “whole sea” of evidence presented to him, whereas an appellate court “will only be island hopping”. As the then Chancellor, Sir Geoffrey Vos, said in Bank St Petersburg at [33] (in a judgment with which Patten and Males LJJ agreed), the danger of island hopping is acute where a determination is made after a long trial. The challenges under grounds 3, 4 and 5 of this appeal provide a stark illustration of that danger, which in my view is further increased when an appellate court is invited to apply an additional level of scrutiny to a delayed judgment.

Secondly, it is not the case that the existence of a material delay gives a disappointed litigant an uninhibited ability to challenge factual findings with which they disagree. As Arden LJ said, the starting point remains whether the judge was plainly wrong. Further, that test applies to inferences and evaluations as well as to findings of primary facts. The additional test to which Arden LJ referred applies to mistaken recollections, or failures to recollect or address material evidence, which have some connection with the delay. Thus, I agree with Ms Demetriou that a causal link is required. In Arden LJ’s formulation the bar is set fairly low (“unless the error could not be due to the delay in the delivery of judgment”). More recently, in Pickle Properties Ltd v Plant [2021] UKPC 6 at [29] Lord Hodge referred to “a basis for believing that there may have been a causal link between the excessive delay and the alleged errors or failings in the judgment”. I do not think there is a material difference between these formulations, but for myself I find Lord Hodge’s approach somewhat easier to apply.

Thirdly, the critical test is whether the appellate court is satisfied that the judge’s decision is safe, or conversely (and as Arden LJ put it in Bond v Dunster Properties) whether it cannot be satisfied that the judge came to the right conclusion. (As to a judgment being “safe” or otherwise, see for example the citation in Bank St Petersburg at [80] from the judgment of Lord Scott in Cobham v Frett [2001] 1 WLR 1775, 1784 that the appellate court “must be satisfied that the judgment is not safe and that to allow it to stand would be unfair to the complainant”.)

It is true that the application of that test does not require an appellate court to disturb factual findings that are (or may be) tainted by delay only if they are “plainly wrong”. It is sufficient if, as in Natwest v Bilta, the documents to which the judge failed to refer “could have made a difference to the outcome” (see at [97] and [121]). However, that does not mean that the test should be reinterpreted, as it appears to me that P4u sought at times to do, as meaning that any evidence which may not have been fully considered is sufficient to call a delayed judgment into question if there is any possibility that it might have made a difference. That, in my view, is too low a bar which would inappropriately open the floodgates for challenges to findings of fact, whatever the materiality or otherwise of the evidence in question, and would be unfair to respondents. It is also not what Natwest v Bilta says.

Rather, faced with a seriously delayed judgment an appellate court must decide whether it is satisfied that the judgment is safe. In order to do this the court cannot confine itself to the islands between which the appellant would wish the court to hop (to adapt Lewison LJ’s metaphor in Fage v Chobani). The court must carefully consider the judgment, and the judge’s consideration of the other evidence, as a whole. As Lord Scott also observed in Cobham v Frett at p. 1783, “excessive delay in delivery of a judgment may require a very careful perusal of the judge’s findings of fact and of his reasons for his conclusions in order to ensure that the delay has not caused injustice to the losing party”. Any other approach would not allow the true materiality (or otherwise) of the evidence on which the appellant relies, and thus the potential for injustice, to be determined. If, on analysis and in the context of the judgment as a whole, that evidence was not material then the court is very likely to be satisfied that the judge came to the right conclusion.

In contrast, in NatWest v Bilta the appellants’ submission, which was accepted, was that the judge in that case overlooked key documents in determining central issues about the dishonesty of the relevant traders and whether they had asked appropriate questions at the “CarbonDesk dinner”, such as to call into question the judge’s conclusion that the traders had turned a blind eye to VAT fraud. The documents in question were material evidence to which the judge had not referred; they were not “minor or peripheral”, and as such the judge’s omission to address them could not be treated as immaterial (see at [118] and [121]).

In this context, it is worth bearing in mind that the nature of an appeal hearing may itself result in the appearance of greater materiality than there in fact is on a close analysis. It is inevitable that an appellant will focus their submissions on the challenges they wish to make. A respondent will obviously need to respond to those challenges and may have limited time to take the court through other evidence that provides the full context. This reinforces the need for the court to pay close attention to the entirety of the judge’s findings.

Fourthly, it is important, in deciding whether any issue raised in relation to delay is material, to consider all such challenges in the round, together with other relevant grounds of appeal that might touch on whether the court can be satisfied that the judgment is, overall, safe. Although, therefore, I have addressed the individual challenges to the judge’s fact-finding in turn, I have also considered them in the round, together with grounds 3, 5 and 7.

The respondents did not seek to maintain that the 15 month delay in this case was not a serious delay, so I will proceed on the basis that the principles considered in NatWest v Bilta and Bank St Petersburg are engaged.

Turning to compartmentalisation, P4u relied on Lord Steyn’s observation in Smith New Court Securities v Citibank [1997] AC 254, 276-277 that a compartmentalised approach may render a judgment unsafe. In fact, the comments made by Lord Steyn relate specifically to findings about the credibility and reliability of witnesses, and the risk of assessing that in isolation from other evidence and inherent probabilities.

The question of compartmentalisation was considered more generally in Bank St Petersburg, and in the context of a severely delayed judgment (in that case, as in Bond v Dunster Properties, 22 months). In Bank St Petersburg the Chancellor observed at [59] that the judge appeared to have rather compartmentalised his treatment of the appellants’ points, and that while his judgment was structured in a logical and comprehensive way it lacked “an element of standing back and considering the effects and implications of the facts he had found taken in the round” (see also at [82], suggesting that different parts of the judgment may have been written at very different times). Whether that was fatal would however depend on “whether it could properly be said that the somewhat piecemeal approach that he adopted unfairly affected the judge’s evaluation of the facts”. The question was whether inconsistencies (examples of which the Chancellor identified) were “sufficient to render the judgment inherently unsafe” ([67]). In that case misgivings expressed by the judge in one part of his (very lengthy) judgment were at least partially at odds with a significant number of his findings of fact, such that the judgment as a whole could not be regarded as safe ([68]).

It can be seen from this that the problem of compartmentalisation is not necessarily linked to delay. It all depends on the judgment in question. In NatWest v Bilta at [85] the court rejected a submission based on the structure of the judgment that the judge had taken a compartmentalised approach. However, in Bank St Petersburg the Chancellor observed at [72] that the delay, as well as the scale of the task, had most likely caused the problems. In that case the judgment must have been, and reads as if it was, prepared in stages with individual findings not quite having the “necessary interaction with what has gone before”. (In contrast, the Chancellor concluded at [81] that the approach in Bond v Dunster Properties had no application in Bank St Petersburg because the judge had carefully re-read all the transcripts.)

In summary therefore, the key question, as with delay, is whether a judgment has been rendered unsafe, but in the context of compartmentalisation the question is whether it is rendered unsafe by a failure properly to consider findings of fact in the round, so that the implications of other findings and evidence are properly evaluated.

In the discussion that follows, references to “NatWest v Bilta delay” and to “compartmentalisation” should be read as shorthand references to the situations where the principles I have just discussed are engaged, such that the court will not allow a judgment to stand unless it can be satisfied that the judge’s decision is safe.

The challenge under this part of ground 4 is that the judge disregarded three items of material evidence in reaching his factual conclusions about the Landmark lunch. These are part of a manuscript note of the initial conversation that Mr Blendis had with Mr Rowe of Slaughter and May on 17 October 2012 (see [33] above), an alleged EE concern reflected in the cross-examination of Mr Blendis that O2 might approach the competition authorities and blame EE, and part of the cross-examination of Mr Swantee. I will deal with these in turn, although I note that (as already discussed) the existence of collusion is in any event firmly countered by EE’s rebuttal of the Anic presumption through the new agreement that EE reached with P4u, and amended to increase volume, in October and December 2012.

The manuscript note is brief. There are some bullet points that obviously summarise Mr Rowe’s initial advice. The judge refers to one of these at [176], namely the “need to document our take”, i.e. that EE needed to record its version of what had happened, but P4u argues that the judge wrongly failed to address another, which reads: “Risk that RD [Mr Dunne] has gone back and taken the conversation as a green light”. In contrast, in his cross-examination Mr Swantee had referred to Mr Dunne not taking Mr Swantee’s reaction at the lunch “as a green light”, a choice of words which P4u says was not coincidental and contradicted the note. Although the judge explained the omission in the permission judgment on the basis that, as a solicitor’s view of possible risks, it was not relevant, the note was a contemporaneous document and recorded an informed factual assessment based on the recording and transcript, which were not available (see [30] above).

I do not consider that there is anything in this point. Read as a whole, the note is clearly based on a preliminary conversation without full details being provided. The first bullet point, which appears immediately before the green light reference, is “What did Olaf [Swantee] say in response to recent phone calls”, indicating that details had not as yet been provided. Although the reference here is to phone calls rather than a lunch meeting, the only relevant discussion with Mr Dunne at the time was that lunch, and other parts of the note also clearly refer to the lunch. If, as P4u maintain is the case, the “green light” reference is to the lunch then the context strongly indicates that the first bullet point extends to the lunch as well.

As the judge found, Mr Blendis took further advice from Mr Rowe in relation to the 25 October email and “script” he prepared. If Mr Rowe had believed that either of these conflicted with what he had previously understood, he would surely have said so. The judge also found that the email reflected both the transcript and the account Mr Swantee had given Mr Blendis shortly after the lunch.

Further, an initial response from a competition specialist that there was a “risk” is hardly surprising. It is fully consistent with the advice reflected in the 25 October email, which refers to risks being “contained and moderated”, not non-existent. But as the judge said, this is irrelevant.

The second piece of evidence is said to illustrate a concern on the part of EE that O2 might approach the competition authorities. P4u relies on responses by Mr Blendis to a question as to whether it would have been easiest to clarify non-engagement by sending a letter of protest. After saying that he was not sure he agreed, because “we didn’t really know what Mr Dunne was trying to say”, Mr Blendis added, “And the response… could have been to escalate and disagree with the − the direction of that information sharing”. When later asked what this meant, Mr Blendis said, “it would have been easy for Mr Dunne to have said that he didn’t share anything and it was Mr Swantee that was imparting information to him”. Given that EE actually had a recording of the discussion at the time, this is said to provide strong evidence that the recording showed that Mr Swantee’s response was sufficient to give Mr Dunne some encouragement, since otherwise there could be no basis for the concern.

I disagree that this is material evidence that the judge was required specifically to address. It was clear from the evidence that at the time there was a general concern, discussed with Slaughter and May, that a proactive response to any of O2, Vodafone and CPW could provoke a defensive report to the competition authorities (see for example the antepenultimate paragraph of the 25 October email set out at [34] above). Such a concern would be a legitimate one even if, ultimately, it could be shown that EE had done nothing wrong. Further, as far as O2 and the Landmark lunch was concerned the iPad recording was incomplete and unclear.

In contrast, at [185] the judge records powerful evidence of a note taken by Mr Blendis of a conversation with Mr Smith, O2’s General Counsel, in December 2012, following the decision in November to escalate EE’s response ([35] above). The note records Mr Smith as saying:

“[Mr Dunne] doesn’t accept he said anything that terrible but acknowledges he asked some questions & posed some scenarios and Olaf didn’t respond. It won’t happen again, and they appreciate the way we’ve dealt with it.” (Emphasis supplied.)

Albeit that it is of a later date, this is also consistent with the reference in Mr Whiting’s email dated 31 January 2014 to Mr Swantee being “not inclined to discuss EE distribution plans” ([46] above), an email the accuracy of which is positively relied on by P4u.

The third item is part of Mr Swantee’s cross-examination, where he accepted that in hindsight it would have been better if he had been clearer, or “even clearer”, with Mr Dunne to avoid a risk of him being encouraged that he could rely on EE not to take up volume with the indirect retailers if O2 exited, but reiterated that he did not think that Mr Dunne had actually got any comfort. P4u relies on this as recognition by Mr Swantee that there was a risk that Mr Dunne might have taken encouragement, and says that this is inconsistent with other parts of his evidence.

Again, there is nothing in this point. Mr Swantee gave his evidence “very frankly” ([36] above). It is both obvious that Mr Swantee could have actively rejected Mr Dunne’s overture and (given the dispute that has arisen) natural for Mr Swantee to observe with the benefit of hindsight that it would have been better if he had taken that course. However, as he reconfirmed in this part of his evidence he did not believe that he had in fact provided any comfort.

P4u maintains that, in failing to find that Mr Dunne and Mr Laurence had a collusive conversation about indirect retailers in 2012 (see [43] above), the judge ignored two crucial documents. These were a slide in the presentation for O2’s off-site meeting on 12 and 13 September 2012 ([26] above) and the text message from Mr Evans dated 2 January 2013 set out at [37] above.

The heading in the relevant slide referred to “events coming up that could create the opportunity to break the current indirect model”. It set out a timeline to 2014 which covered the launch of EE’s 4G offering, expiry dates for O2’s contracts with P4u and references to possible changes to CPW’s then joint venture structure. Stated assumptions included that if O2 withdrew from an indirect retailer then another MNO would follow. The main text in the slide, under the heading “Business Impact”, referred to uncertainties over the impact of 4G, O2’s then proposal for a short-term deal with P4u (see [28] above), the possible opportunity created by a change to CPW’s structure and O2’s “handset solution” (a handset financing proposal that would be available only to direct customers of O2, judgment at [130]). The final point reads as follows:

“O2 UK withdraw from P4U in January 2014 from both upgrades and gross as adequate volumes are being achieved through the direct channels allowing for greater bargaining power with remaining direct (assumes Vodafone follows suit) and existing P4U assets could be purchased for lower MV than current price.”

P4u relies on the fact that this refers only to Vodafone as indicating that O2 knew enough to consider that its most likely response to an O2 exit would be to follow suit. In contrast, in relation to the text message, P4u relies on the absence of any reference to Vodafone, the implication being that (unlike Mr Swantee at EE) Vodafone was already on board in its “resolve against the Indirects”.

Mr MacLean stressed that not only had the judge disregarded this evidence, on which it was clear that P4u placed significant reliance, but the significance of these documents was all the greater because the inherent probabilities all pointed one way. Mr Dunne was obviously prepared to collude, he had a much friendlier relationship with Mr Laurence than with Mr Swantee, and he had the same incentive to “de-risk” O2’s position with Vodafone as with EE.

Further (and a point on which ground 5 was also engaged) 4G pricing was, Mr MacLean submitted, a closely related topic. The problematic conversation at the Landmark lunch started with 4G pricing and identified the particular risk of discounting by indirect retailers: it was a composite concern. The judge had failed properly to take account of the relevance of the coordination that he had found to exist between Mr Dunne and Mr Laurence over 4G pricing. Rather, he considered that they were distinct topics.

Although P4u clearly placed reliance on them at trial, I am not persuaded that the slide and text render unsafe the judge’s conclusion that there was insufficient evidence to find that there was collusion between Mr Dunne and Mr Laurence over indirect retailers. The text message obviously provides no positive evidence of collusion with Vodafone, since it is not mentioned. The slide is part of a lengthy presentation to which the judge referred in some detail in describing O2’s decision-making at [129]-[131]. The assumption referred to is just that, and could have been made for a variety of reasons (about which there was speculation in cross-examination) other than that it reflected information from Vodafone. Further, the presentation actually recommended a further one year contract with P4u, rather than the provisional decision to exit that emerged from the meeting. In addition, to the extent – as appears to be the case – that P4u is suggesting that Mr Dunne and Mr Laurence colluded about indirect retailers after that decision was reached with a view to “de-risking” it (for example, at their meeting on 10 October), the presentation predated both events.

Further, while I accept that a challenge on grounds of NatWest v Bilta delay is not precluded simply because the judge has referred to the evidence in another context, it is not irrelevant that the judge turned his mind to both documents in addressing the closely-related question of whether Mr Dunne colluded with Mr Swantee. That calls into question whether the judge’s failure to address them explicitly in the context of the allegation of collusion between Mr Dunne and Mr Laurence was anything to do with the delay in producing the judgment, rather than a decision not to lengthen it further and instead to address the point more generally with his comment at [241] that there was a “wholly insufficient basis on the evidence to support such a finding” ([43] above).

In addition, and critically, the judge found that Vodafone had developed no strategy regarding either P4u or CPW in September 2012 (the date of the slide), and further that it only started developing its strategy from November 2013, following which there were detailed negotiations with P4u for an extension to their agreement and, later, discussions for a new deal with CPW (see [50], [55], [56] and [58] above). This is wholly inconsistent with any plan to “follow suit” or having “resolve against the Indirects” (if that meant withdrawing or reducing volume rather than anything else) in September 2012 and January 2013. On a proper analysis, therefore, the slide and text do not provide material evidence that calls into question the judge’s conclusion.

As regards ground 5, I do not consider that there is any real risk of compartmentalisation. The judge was well aware of how the conversation at the Landmark lunch developed from 4G pricing to the position of indirect retailers, and made careful findings about it, rejecting Mr Dunne’s evidence that he did not seek to influence EE’s pricing strategy and explicitly finding that he had said that discounting was likely to be led by the indirect retailers (judgment at [195]). But the judge was also well aware of the separate approaches made shortly afterwards to Mr Swantee by Mr Laurence and Mr Harrison of CPW, and Mr Dunne’s involvement in that. Indeed (and for what it is worth), he dealt with those approaches as part of the same discussion as the Landmark lunch, in section G of the judgment, and the specific allegation about Mr Dunne and Mr Laurence in the immediately following section. Those later approaches to Mr Swantee were specifically about 4G pricing and, as Mr Harrison’s involvement rather emphasises, did not encompass steps against indirect retailers. The two topics were not therefore necessarily linked. This is nothing like the compartmentalisation found in Bank St Petersburg.

P4u maintains that, in concluding that there was no forward-looking discussion between Mr Humm and Ms Castillo in September 2013 ([49] above), the judge omitted to deal with an email that was at the centre of its case and featured heavily in its closing submissions. This was an internal email sent by Ms Rose of Vodafone on 4 November 2013 to, among others, Mr Laurence and Mr Hoencamp. After a reference to P4u claiming to be in negotiations with O2 “to either bring them fully back in or to fully kick them out” Ms Rose added:

“Our intel suggests that O2 is more likely to align with CPW and come out of P4U completely”.

Mr MacLean submitted that this demonstrates that Ms Castillo did provide forward-looking information at the meeting a few weeks before, contrary to the judge’s conclusion. Further, even though Vodafone only reached its own decision to exit P4u later (and after O2’s own exit), understanding O2’s attitude remained material for it due to the risk that O2 would re-enter a relationship with P4u.

Mr MacLean also referred to Ms Rose’s statement in cross-examination that she prided herself on her integrity, and submitted that the judge should have considered that alongside the fact that she was involved in the subterfuge aimed at discouraging P4u from pursuing an alternative strategy with BT (see [61] above). He submitted that this made clear that she was prepared to deceive, and in the present context the judge should have drawn the obvious conclusion that the information came from Ms Castillo and was passed on by Mr Humm.

P4u also maintains that, relevant to the judge’s reliance on the reciprocity of any exchange (see [48] and [51] above), two further pieces of evidence undermine the suggestions that Mr Humm disclosed nothing to Ms Castillo and that Ms Castillo would be unwilling to participate in any unilateral disclosure. The first is evidence suggesting that O2 may have received information about Vodafone’s contract with CPW, which it is said could have been provided as part of a reciprocal exchange at the meeting. It is a manuscript note by a different Telefonica executive which reads “VOD-CPWH  no tail”. The second relates, in essence, to Ms Castillo’s probity and the judge’s findings about a lax attitude of senior executives to competition law (see [24] above). P4u relies on some brief notes of a call on 7 October 2013 between Ms Castillo and Ms Claudia Nemat, DT’s representative on the EE board, recording (in translation) “BT – we do it in reasonable conditions”. This is said to relate to an agreement that EE was in negotiation with BT about, for BT to use EE’s network infrastructure. Ms Castillo could not explain the note.

It is important to focus on the judge’s actual findings. He concluded that Ms Castillo had not been briefed to discuss the topic and that what Mr Humm actually asked was how Ms Castillo felt that O2 was doing. Further, Mr Humm could not give any commitment or comfort because at the time Vodafone had not even begun to address its own position with indirect retailers. And Telefonica had at that stage also not yet been convinced that O2 should exit P4u.

The suggestion that a reciprocal exchange was supported by O2 receiving from Mr Humm a small piece of information about Vodafone’s contract with CPW (which was not developed by Mr MacLean in oral submissions before us and which I understand was not put to Mr Humm in cross-examination at the trial) is wholly speculative and cannot realistically be regarded as a material omission. The executive who made the note, a Mr Valera Sanchez, whom the judge found to be a frank and honest witness (judgment [122]), was cross-examined on the document and, in response to a question from the judge, suggested that he would expect this sort of information to come from discussions with indirect retailers.

I also cannot see anything in the exchange with Ms Nemat, even assuming that the (rather anodyne) information regarding BT was confidential. The judge’s findings were not based on Ms Castillo’s attitude to competition law or her willingness to receive confidential information from a competitor. Rather, the judge found that as a “highly accomplished business person” Ms Castillo would not have given a commitment to a competitor as to O2’s future conduct or disclosed its internal strategy (judgment at [291]). Further, although O2 had by 9 September decided that it should exit P4u that was not approved by Telefonica until late October ([45] above). In that context, the suggestion now made that Ms Castillo went even further and gave comfort that O2 did not intend to re-enter a relationship with P4u lacks reality, quite apart from the fact that even if she had been in a position to do so Mr Humm had nothing of substance to give in return.

That leaves Ms Rose’s email. As Mr Williams recognised, it would certainly have been preferable if the judge had addressed it. It was obviously raised in closing submissions, which (based on references in both the judgment and permission judgment) the judge must have re-read. It seems highly unlikely that the judge then forgot about it when making his decision about the meeting between Mr Humm and Ms Castillo. This is reinforced by the permission judgment, which states that the email was not taken into account because it was “not relevant”, since the change in Vodafone’s thinking about P4u occurred much later, and after O2’s exit. If that is right, then P4u’s criticism of that approach is to no avail. That is because the omission would have nothing to do with delay (so that the relevant test on appeal is confined to whether the judge was plainly wrong in his approach, see above), but more pertinently the issue would fall outside the grounds of appeal altogether given that ground 4 (delay) is the only relevant ground.

I have reservations about relying only on what the judge said in his permission judgment, albeit that I have no reason to doubt its accuracy. A permission judgment can be considered in order to elucidate the full reasons for a judge’s decision: see [122] above. But there is a natural risk of there being at least a perception that a statement about why something was omitted from a judgment is an ex post facto explanation, rather than part of the judge’s actual reasoning at the time.

However, in this case there are other considerations. First, it is in fact unlikely that the judge did omit consideration of the email given its prominence, and also the obvious care he took. That care is obvious from the judgment and also from the fact that, as he explained in the permission judgment at [39], he re-read all the documents referred to during the course of the trial (in addition to many additional documents footnoted in closing submissions). That would have included the email from Ms Rose.

Secondly, it was clearly correct that Mr Humm had no substantive information to convey about Vodafone’s position and that O2’s position was not settled. I have also concluded that the judge’s finding that Ms Castillo would not in any event have volunteered information by way of a one-way exchange cannot be impugned. The existence of Ms Rose’s email cannot affect those important points, which make collusion inherently unlikely.

Thirdly, a careful consideration of the email as a whole helps to put the comment on which P4u relies in context. Rather than indicating collusion between O2 and Vodafone over exit from P4u, it raises two options for Vodafone doing continued business with P4u. These were, first, a longer term “tied model” proposal (which I understand would have involved Vodafone taking a minority equity stake in P4u) or, secondly, a two year extension to the existing contract which would “lock in P4U now” while Vodafone’s longer term direct distribution strategy was executed and also do what P4u required from Vodafone to “kick O2 out completely”. These alternatives and references are hardly indicative of collusion, rather the opposite.

Fourthly, although Ms Rose’s reference to “intel” does have the flavour of something going beyond pure market information or press speculation, there is obviously no indication as to its source. The fact that it may not have been directly from O2 is rather illustrated by a comment in the email, immediately after the words on which P4u relies, that “I think P4U know this”.

Fifthly, the email provides no support at all for the argument that the significance of the exchange between Ms Castillo and Mr Humm was that it went beyond indicating that O2 would exit P4u (a matter that became public well before Vodafone’s own decision, see above) and provided any degree of comfort to Vodafone that O2 would not re-enter a relationship with P4u. Vodafone’s continued uncertainty about O2’s position for many months after the meeting is evident from the documentary evidence, contrary to the suggestion that it was materially reduced.

Sixthly, I do not consider that the challenge to Ms Rose’s integrity is material. The email can be assessed for what it says, not by reference to Ms Rose’s evidence.

Finally, the judge concluded that even if there had been a relevant exchange of information then the Anic presumption was rebutted by Vodafone’s later conduct, including signing an overlay agreement with P4u: see [49] above.

In its supplementary skeleton argument P4u claimed that the judge’s conclusion at [292] that the Anic presumption was rebutted related to his finding that Ms Castillo had not shed light on future strategy rather than to the alternative that, contrary to his finding, she had done so. However, that is not the correct reading of that paragraph, the relevant part of which reads:

“However, since this does not involve disclosure of future strategy I do not think it supports even the most general allegation of anti-competitive conduct pleaded at para 131(b)(iv). But even if that occurred (and I emphasise that I make no finding since that would be pure speculation) [i.e. even if future strategy was disclosed, which is what para 131(b)(iv) concerned], I consider that such unilateral passing of information by Telefónica Europe to Vodafone could have no causative effect on O2’s own conduct. And I find that any presumption that it had an effect on Vodafone’s conduct is rebutted by the fact that Vodafone UK’s decision to leave P4u was not reached until many months later, in mid-2014, after significant developments in the UK market and extensive and detailed negotiation between Vodafone UK and P4u.” (emphasis supplied).

Ultimately, therefore, any challenge based on Ms Rose’s email can go nowhere.

This challenge relates to the email from Mr Whiting set out at [46] above, in which Mr Whiting recorded what he said he had been told by Mr Dunne on 27 January 2014. P4u claims that, when deciding that Mr Dunne had made up the “commitments” given to Mr Colao and Mr Humm, the judge had disregarded two important pieces of evidence. The first was that what Mr Dunne said about Mr Swantee was true, and the second was the extraordinary coincidence of the reference to Ms Castillo and Mr Humm, given that the subject of O2’s exit had come up at their only meeting. These points should have led the judge to consider whether his conclusion that Mr Dunne was lying was improbable.

I do not consider that these points are properly within the scope of the complaints about either NatWest v Bilta delay under ground 4 or compartmentalisation under ground 5. The judge clearly did not fail to recall or address what was said about Mr Swantee; indeed he relied on it at [307d)] in deciding that Mr Dunne did say what was attributed to him. He also dealt with the meeting between Ms Castillo and Mr Humm, along with the allegation of collusion between Mr Alierta and Mr Colao, in the very section of the judgment that addressed the email from Mr Whiting. I do not consider that the judge can fairly be said to have failed to consider both the email and other evidence of collusion between those individuals in the round. Indeed, he explicitly did so at [305] and [311]. As he said at [311]:

“I have considered carefully what impact my finding that the conversation indeed took place has on the assessment of whether what Mr Dunne said was true. In my judgment, it does not affect that determination. All the reasons remain for holding that the alleged collusion between O2/Telefónica and Vodafone is most unlikely; and, as Mr Lloyd pointed out at the time, it is just as curious that Mr Dunne should have made these remarks if they were true as if they were not.”

(The reference to Mr Lloyd, P4u’s Chief Legal Officer, is to his evidence that it had been difficult to believe that Mr Dunne would have made the statements had they been true; it would be “like robbing a bank and telling everyone you have done it”: judgment at [308].)

Mr Williams also relied on the fact that the allegation of a commitment to Mr Colao was clearly false, weakening P4u’s reliance on the truth of what was said about Mr Swantee as indicative of all Mr Dunne’s statements to Mr Whiting being true. That allegation had not been put to Mr Colao in cross-examination, had “virtually collapsed” by the end of the trial and was not challenged on appeal. Mr MacLean disagreed that P4u had dropped the allegation, but nevertheless the strength of the judge’s findings about the absence of collusion between Mr Alierta and Mr Colao (and therefore the falsity of one part of what Mr Dunne said) weakens P4u’s reliance on the truth of the comment about Mr Swantee.

P4u maintains that there are three key areas in which the judge disregarded or mis-recollected important evidence in concluding that there was no collusion between Vodafone and EE in 2014, namely 1) his conclusion that the topic Mr Scheen wished to raise with Mr Humm was a joint acquisition; 2) his finding that the change in EE’s modelling did not reflect a new view that P4u would go out of business; and 3) his finding that the failure to minute the relevant discussion at EE’s May 2014 BRM was not suspicious. In relation to the missing minutes, P4u also makes a compartmentalisation challenge about Mr Harris’s involvement in misleading Ofcom.

In concluding that the topic Mr Scheen wished to raise was a joint acquisition, the judge took account of Mr Scheen’s evidence that, despite being concerned about the cost, he had thought that a joint acquisition was “worthy of further investigation” (judgment at [465] and [573]). Mr MacLean submitted that the judge wrongly disregarded other parts of Mr Scheen’s evidence which made his scepticism about that idea clear, an email sent to Mr Scheen two days before the call with Mr Humm by Mr Naulleau (also of Orange) and witness evidence from Mr Naulleau and from Mr Thomas Dannenfeldt, DT’s CFO and a member of EE’s board. That evidence pointed to an overwhelming likelihood that the topic that Mr Scheen had wished to raise, and did raise, was exiting P4u. Further, just raising that topic, even at a very high level, would fall foul of Article 101, so it was no answer that the conversation was brief.

Ms Demetriou submitted that there was an ample basis for the judge’s conclusion. Not only was a joint acquisition under active consideration at the time but neither EE nor Vodafone had reached their own decision as to whether to stay with or leave P4u. The judge found at [574] that other EE directors were not told about the discussion, EE’s modelling continued on the premise that Vodafone would remain in P4u and, at the May 2014 BRM, there was both no firm decision to leave P4u and the idea of an acquisition also resurfaced. Further, P4u mischaracterised the evidence, which was also taken into account by the judge alongside other evidence.

I have carefully reviewed the evidence relied on by P4u and do not consider that it materially supports P4u’s case.

It is true that Mr Scheen expressed scepticism about an acquisition (because in his experience such transactions were expensive, time consuming and risky) but it does not follow that he would not have considered it at all, alongside other possibilities. Indeed, unless an acquisition was clearly a non-starter a failure to consider it would hardly be consistent with sound business decision-making. But it was obviously not a non-starter. The idea was being promoted by EE and was to be discussed again at the BRM on 10 April (see [70] above). The shareholders had also agreed at the March BRM that their own mergers and acquisitions teams should consider the possibility of an acquisition, which duly occurred (judgment at [452] and following). That possibility was then duly discussed again at the April BRM and was not ruled out. Although the minutes of the March BRM also record Mr Scheen as commenting that “if [the CPW/Dixons] merger goes ahead, may make sense to exit P4U” (a comment which was noted by the judge at [449]), that is a statement that is obviously in highly conditional terms.

The witness evidence of Mr Naulleau and Mr Dannenfeldt reflects what was at the very least a distinct lack of enthusiasm for an acquisition on the part of Orange in particular, and notably Ms Lambert’s strong opposition, but that opposition was recorded by the judge at [463]. Mr Scheen’s own reservations are recorded at [465]. Again, the evidence does not detract from the fact that the idea of an acquisition was nevertheless under active consideration.

As to the email to Mr Scheen from Mr Naulleau, this forwarded an earlier email from Mr Naulleau to Mr Gervais Pellissier, Orange’s CFO, which contained a briefing for a call that Mr Pellissier was to have with Mr Dannenfeldt. The judge referred to the forwarded email at [464] and [467], so he had clearly considered it. That email referred to the “unique opportunity” offered by the announcement of the CPW/Dixons merger “to break EE’s extreme reliance on UK indirect distribution”, and then made a series of points. It noted that EE had a “window and lever” depending on whether or not it chose to support the merger. One “scenario for EE to explore” was to negotiate support for the merger in exchange for an end to RMS (meaning revenue margin share, an ongoing monthly payment by MNOs to indirect retailers: judgment at [432]) and withdrawing from P4u. But EE was “more inclined to analyse” an acquisition of P4u. The email also noted that Mr Dannenfeldt’s position was “first and foremost to do everything to weaken the excessive weight of indirect distribution in the UK” and that Ms Lambert did not support the acquisition scenario and favoured moving forward on a “costed analysis of our more breakaway scenario of stopping RMS with CPWH and withdrawing [P4u]”.

The content of this email is best understood in the light of the fact that, as Ms Demetriou confirmed based on Mr Naulleau’s first witness statement, EE’s contract with CPW had a change of control clause. The engagement of that clause on a merger, with the effect that EE would have been entitled to terminate the contract early, provided EE with the opportunity to renegotiate and improve the financial terms it had previously agreed with CPW. If EE could achieve that then its contract with P4u would obviously be correspondingly less attractive. (It is worth noting that even the apparently implacably opposed Ms Lambert linked withdrawing from P4u with managing to get improved terms from CPW, which at this stage would have been far from assured since the merger terms were not even agreed: see [53] above.)

Thus, the “opportunity” was the one provided by the CPW/Dixons merger. That gave rise to options that included leaving P4u if improved terms could be agreed with CPW, or acquiring it. As Mr Naulleau also explained in cross-examination and is reflected in an email of 9 April summarised by the judge at [469], another alternative would have been to use the “game changer” of the merger to insist on improved terms from a comparatively weakened P4u. Mr Naulleau’s email is simply not consistent with the “opportunity” Mr Scheen raised with Mr Humm being more likely to be one of leaving P4u rather than anything else.

The judge summarised his assessment of the attitude of EE’s shareholders to a joint acquisition at [475]. Based on the evidence we were shown that assessment seems to me to be entirely fair. I am not persuaded that there was any mis-recollection or failure to address material evidence such as to amount to Natwest v Bilta delay.

That conclusion makes it strictly unnecessary to deal with a further argument by Ms Demetriou and Mr Williams that, in any event (and beyond referring to the potential for change in the UK market created by the proposed merger), the judge found that no conversation of substance – whether about a joint acquisition or otherwise – took place: see [89] above. In brief, however, there is real strength in that point, which is also reflected in the permission judgment. The 5 minute call on which P4u relies was on an open line (indeed, in cross-examination Mr Humm explained that he was in a public space, at an airport) between individuals who did not know each other, unlike the “secured” call that Mr Scheen then wanted to arrange.

Further, and as Mr Williams submitted, the reference in one of the texts to a “joint opportunity” ([87] above) is consistent with the topic being a joint acquisition. And even if that is not right it is, at the least, consistent with the fact that the “opportunity” that had arisen was the one that was created by the CPW/Dixons merger, as discussed in Mr Naulleau’s email.

Mr MacLean made something of the fact that the discussion of the exchanges between Mr Scheen and Mr Humm is in a separate section of the judgment, rather than in the judge’s chronological account. To the extent that is a complaint about compartmentalisation, I do not consider that there is anything in the point, and it also fails to reflect other critical, relevant, findings. It was up to the judge to decide how to structure his judgment, and the level of cross-referencing shows that the judge had all the relevant issues well in mind.

First, the earlier discussion about Vodafone’s decision-making has a specific section about Mr Humm which cross-refers to the judge’s consideration of the exchanges between Mr Scheen and Mr Humm (see [66] above). Further, in that discussion the judge made clear that if P4u was right then Mr Humm must have kept the information he obtained from Mr Scheen to himself for months rather than sharing it with Vodafone UK, as he said he would have done. That is highly improbable.

Secondly, the section of the judgment dealing with exchanges between Mr Scheen and Mr Humm cross-refers at [576] to the earlier findings that, at that stage, neither Vodafone nor EE had agreed their strategy, and that EE continued to assume that Vodafone UK would remain with P4u: see [89] above. These are further, powerful, pointers away from any collusion in the conversation on 2 April.

P4u relies on EE’s modelling of what it expected to happen if it left P4u as having changed dramatically on 5 May 2014. Mr Eyre went from assuming that EE would retain only 25% of its P4u volumes to an absorption rate of 60-70%, and also assumed handset cost efficiencies of £34m, an assumption that had previously been made in relation to a scenario in which P4u was acquired. (The absorption rate was the percentage of sales volume through the retailer assumed to be retained if EE ceased to deal with it. “Handset cost efficiencies” is a euphemism for higher margins through lower handset subsidies.)

P4u’s case was that these changes followed Mr Eyre’s discussions with Mr Milsom and Mr Allera on 28 and 29 April ([71] above), and further that in concluding that the 5 May slides did not assume that P4u would collapse the judge disregarded Mr Eyre’s witness statement. Mr Eyre’s evidence was that the assumption about handset cost efficiencies reflected a view that P4u was likely to represent at least a significantly reduced presence in the market and that the increased absorption rate also assumed that P4u was more likely to cease trading following EE’s withdrawal. Further, in cross-examination Mr Eyre had confirmed that at this point in the modelling process he had started to assume that P4u would “materially exit the market”. The judge had not addressed why these assumptions were adopted on 5 May, and if he had he would have had to agree that it was very likely to have been his discussions with Mr Milsom and Mr Allera. Instead the judge wrongly attributed the assumption that P4u would collapse to Mr Eyre’s later analysis of P4u’s cash position ([76] above).

Mr Eyre’s witness statement makes a number of points. These include the iterative nature of the modelling process as EE’s thinking evolved, with interim updates not necessarily being fully thought through. It makes clear that Mr Eyre could not recall the precise rationale for each change in assumptions. As to the 5 May version, the witness statement refers to it being possible that the absorption rate in those slides reflected a provisional view that P4u “were more likely to cease trading following a withdrawal by EE than to continue trading”, a view which would not have been a particularly considered one at that stage. As to the handset cost efficiencies, Mr Eyre could not recall but it might have reflected a provisional view that P4u was likely to “represent a significantly reduced presence in the market following a withdrawal by EE (because EE could not otherwise, in my view, have realised handset cost efficiencies of this scale)”. These observations do not fit well with P4u’s depiction of a dramatic change in approach on 5 May. It is also clear that Mr Eyre did not have a clear recollection of each iteration. As the judge recorded at [519], Mr Eyre was “doing his best to reconstruct his thinking from the contemporary documents”. The judge was obviously entitled to form his own view of the documentary evidence.

Leaving both this point and the question of whether this is properly a challenge based on NatWest v Bilta delay to one side, there is in any event both a straightforward and a slightly more involved answer to P4u’s complaint about the modelling. The straightforward answer is that the modelling does not support any allegation of collusion with Vodafone over exiting P4u because it consistently assumed that Vodafone would remain with P4u past the expiry of its own contract in October 2014, and indeed after EE’s own exit in September 2015. So even if it could be linked to delay (which I consider to be very doubtful) the alleged failure by the judge to address Mr Eyre’s evidence is immaterial.

The more involved answer is really an embellishment of this. As always, it is important to consider the judgment carefully. The judge conducted an in-depth analysis of the modelling. He recorded P4u’s argument about the change in the assumptions reflected in the 5 May version at [495] but also said that both Mr Eyre and Mrs Derbyshire had been told that changes in absorption rates reflected what was under discussion with CPW (the option in question assumed an enlarged contract with CPW alongside exit from P4u). However, the slides still showed that that option was relatively unattractive and, importantly, it assumed that EE pulled out of P4u at the end of its contract, in September 2015. As the judge observed, therefore, an early collapse of P4u was certainly not assumed.

The slides produced for 7 May contained similar assumptions. The judge found at [498] that the 30% of new acquisitions and 40% of upgrades that it was (conversely) assumed would not be absorbed were assumed to go to competitors, and that those assumptions “were made on the basis that P4u would survive and be able to divert that volume to others”. The judge noted an express statement in a slide in relation to new acquisitions that it was assumed that P4u would be “able to connect the remaining 30% on another MNO”. On this basis the judge was clearly entitled to conclude, as he did at [499], that rather than assuming that P4u would go out of business if EE left, the modelling assumed the opposite. The judge also found that further material changes were made to the assumptions thereafter, including (by 12 May) that, via a combination of a revised deal with CPW and EE’s direct sales channel, 70% of new connections would be absorbed as early as the third quarter of 2016 (a point emphasised by the judge at [506]). These changes had a substantial positive impact on the relative financial attractiveness of leaving P4u.

Mr Eyre’s analysis of P4u’s likely cash position followed direction from shareholder representatives at the workshop on 7 May ([75] above). This showed that P4u would run out of cash in mid-2016 if it failed to find a “second player” following EE’s withdrawal (judgment at [503], quoting the covering email sent by Mr Eyre on 8 May with his analysis of P4u’s cash position, emphasis supplied). This necessarily assumes that Vodafone remained with P4u when EE exited. The point was that P4u would need to replace EE, not both EE and Vodafone. This is further emphasised by the analysis attached to the email, which states as one of its assumptions: “Additional volume from Voda at only 300…”. The judge found at [503] that this was an assumption that Vodafone UK would take an additional 10% of new connections secured by P4u.

The judge gave detailed reasons at [522] to support his conclusion that EE’s modelling did not reflect an assumption that Vodafone UK was likely to pull out of P4u at around the same time as EE, reflecting his earlier findings. These included that the cash flow analysis assumed that Vodafone UK both remained with P4u and increased volumes following EE’s exit, and the contemplation that P4u could survive if it attracted a replacement for EE. In addition, the handset cost efficiencies calculation assumed that for the first 18 months after departure from P4u in September 2015 EE would need to increase its subsidy before later reducing it. As the judge said at [522d)]:

“I regard this projection as inconsistent with an exit by Vodafone UK as well as EE from P4u in 2015, which would obviously have precipitated an immediate collapse of P4u (as indeed occurred).”

The judge was obviously aware of the fact that the assumptions changed in the 5 May version of the slides. But it was only when further work was done after the workshop that Mr Eyre focused on the impact of EE’s exit on P4u’s cash position. This is hardly surprising given the iterative nature of the process, the narrowing of options being considered and the express request to consider what would happen to P4u, a question that had been raised but left for further consideration in the 7 May version of the slides ([74] above). But the critical point is that there is nothing that supports any suggestion of a coordinated exit by EE and Vodafone.

Further points could be added to this. For example, the email Mr Eyre sent to his team on 28 April after his discussions with Mr Milsom and Mr Allera referred to a focus on two scenarios, one in which P4u stopped trading and one in which it continued. That is not easy to reconcile with P4u’s case as to what was said in those discussions. The gradual development of Mr Eyre’s thinking is also illustrated by a further email he sent to Mr Allera on 3 May, in which he observed that he kept coming back to some questions, one of which was “How long do we believe P4U will survive if we pull out”, but attached a slide deck which referred to it being likely to survive (judgment at [520]). However, these additional points do not detract from the key point, which was that, even if EE did pull out, Vodafone was assumed to remain with P4u.

Last but by no means least, the challenge in relation to EE’s modelling was that the judge failed to deal with Mr Eyre’s evidence, such as to engage NatWest v Bilta. I am entirely unpersuaded that the judge did fail properly to address that evidence.

P4u claims that, in deciding that there was no deliberate decision to avoid recording a discussion ([91] above), the judge wrongly failed to address evidence as to past practice in completing Board minutes and “similar-fact-pattern” evidence in relation to Mr Harris. Neither Mr Harris nor Mr Milsom had been able to recall any other failure to complete minutes. Further, in relation to the March BRM Mr Harris had emailed colleagues to say he was checking the minutes with Mr Blendis (who had not been there) because he was being “extra cautious” given the “topic above”. The email was entitled “indirect, next steps”. In cross-examination Mr Harris speculated that he may well have asked Mr Blendis whether “there was a point that had been inappropriately captured from a competition law point of view”.

P4u also maintains that the judge took a compartmentalised approach in relation to EE’s contact with Ofcom, pointing to the judge’s conclusion that it did not lead him to change the conclusion he had “otherwise reached”. Mr Harris’s willingness to mislead EE’s regulator must have a bearing on his motivation for taking a more modest step of not minuting a discussion.

The judge addressed P4u’s submission that there was a deliberate omission from the minutes. He noted Mr Harris’s normal practice in producing and circulating draft minutes and that he could not explain the omission, but also that there was no record of him being chased for them (judgment at [591] and [592]). As noted at [79] above the judge relied on two other contemporaneous documents to make his findings about what had happened, a note by Mr Deloison and an email from Mr Allera to Mr Harris. The judge found at [594] that the email from Mr Allera, which started “As promised” and set out Mr Allera’s understanding of what was agreed, would not have been requested if Mr Harris had been able to take a clear note. I would add to this that it would be surprising if that request had been made if Mr Harris had concluded that he should not minute the discussion at all.

The judge also commented that if there were aspects that there was a preference to avoid minuting it would not be the case that the entire item would be left blank (alongside, here, the AOB section, a heading under which there was also significant discussion) (judgment at [595]). Further Mr Deloison’s note recorded acceptable discussion, except for one reference to “be careful with VOD (anti trust risk)” which the judge found to relate to the joint acquisition possibility ([596]).

I agree that, if Mr Harris had been concerned about not minuting an inappropriate discussion, he would in all likelihood have included something selective and suitably anodyne under that item of the minutes, rather than leaving them obviously blank in a way that only invites suspicion. Mr Allera’s summary could have done that job. The fact that he had obviously requested Mr Allera’s email also suggests that Mr Harris had planned to include something in the minutes.

The judge was not obliged to deal with every item of evidence. The failure to recollect other occasions is far less remarkable than the oddity of a deliberate decision to leave whole sections blank. The judge addressed Mr Harris’s normal practice and was clearly entitled to conclude that something had gone awry on this occasion. Further, Mr Harris’s care in checking with Mr Blendis on a previous occasion would also be far more consistent with including something inoffensive in the minutes rather than nothing at all. I am far from persuaded that any failure to deal with the evidence on which P4u relies is linked to delay, but in any event it is immaterial.

In relation to Ofcom, EE pointed to the fact that Mr Harris had asked colleagues to check the March 2014 date that was wrongly given to Ofcom as the time when the strategy of having CPW as the principal partner was first agreed (rather than May 2014 which the judge found was the correct date). Further, and importantly, I agree with EE that the judge’s choice of words about the conclusion he “otherwise reached” does not indicate compartmentalisation, rather the reverse. The judge explicitly asked himself whether the incident with Ofcom affected his conclusion.

Ground 7 relates to an alleged error in the judge’s approach to Telefonica’s failure to adopt appropriate document preservation measures in 2015, when the allegations were first made, or for some time thereafter. As summarised at [23] and [50] above, the judge applied the approach in Efobi but failed to draw an adverse inference. This was relevant to the allegation of collusion between Mr Alierta and Mr Colao, but this ground of appeal is not limited to that.

At the hearing of the appeal Mr Draper, who made submissions for P4u on this ground, accepted that although Lord Leggatt’s comments in Efobi concerned absent witnesses, it is also correct to apply them to an absence of documentary evidence, as Lewison LJ recognised in Volpi v Volpi [2022] EWCA Civ 464, [2022] 4 WLR 48.

What Lewison LJ said in Volpi v Volpi was this:

“5.

Tribunals are free to draw, or to decline to draw, inferences from the facts of the case before them using their common sense. Whether any positive significance should be attached to the fact that a person has not given evidence, or to the lack of contemporaneous documentation, depends entirely on the context and particular circumstances: Efobi v Royal Mail Group Ltd [2021] UKSC 33; [2021] 1 WLR 3863.”

This summarised the approach taken by Lord Leggatt in Efobi at [41]. There Lord Leggatt emphasised the risk of “making overly legal and technical what really is or ought to be just a matter of ordinary rationality”, and said that so far as possible tribunals should be free to use their common sense in deciding whether or not to draw inferences from the facts “without the need to consult law books”. Specifically in relation to absent witnesses, the tribunal’s approach would depend entirely on the context and circumstances. Relevant considerations would include whether the witness was available, what relevant evidence they might have given, its significance and other relevant evidence that was available.

It follows that an appeal on the ground that an adverse inference should (or should not) have been drawn will generally require the appellant to demonstrate that no reasonable tribunal could have reached the decision it did(Efobi at [42]). That is a high hurdle.

A further point worth stressing relates to other relevant evidence. As Morritt LJ observed in Malhotra v Dhawan [1997] 8 Med LR 319, 323, an adverse inference must be consistent with other evidence and factual findings. Earlier, at p.322, he had described limits on the application of the principle as follows:

“First, if it is found that the destruction of the evidence was carried out deliberately so to as hinder the proof of the plaintiff’s claim, then such finding will obviously reflect on the credibility of the destroyer. In such circumstances it would enable the Court to disregard the evidence of the destroyer in the application of the principle...

Second, if the Court has difficulty in deciding which party’s evidence to accept, then it would be legitimate to resolve that doubt by the application of the presumption. But, thirdly, if the judge forms a clear view, having borne in mind all the difficulties which may arise from the unavailability of material documents, as to which side is telling the truth, I do not accept that the application of the presumption can require the judge to accept evidence he does not believe or to reject evidence he finds to be truthful.”

All of these points are relevant here, but the third makes a particularly pertinent point about the significance of other evidence.

Mr Draper stressed that the failure to adopt document preservation measures was the result of a deliberate decision on the part of Telefonica. The only explanation given was its view that the claim was unmeritorious, such that it was disproportionate and unnecessary to spend time preserving documents for a claim that was highly unlikely to proceed. Telefonica had called no witness who could be cross-examined about that. By proceeding with reference to the personal position of Mr Alierta (who it seems was only informed of the claim much later) the judge wrongly treated the failure as an innocent one, rather than a deliberate decision by the relevant corporate defendants which allowed potentially damaging documents to be lost. Mr Draper also stressed that in cases involving collusion documentary evidence is likely to be sparse and fragmentary, making preservation of whatever documents there are potentially all the more important. Further, the very marked difference in the level of disclosure by Telefonica relative to other defendants showed that there must have been a significant loss of documents. And the judge’s reliance at [303] on the fact that communications by Mr Alierta with O2 or Vodafone would have been captured in those defendants’ disclosure exercises did not take account of the potential significance of internal communications. The judge should have drawn adverse inferences, or as a minimum not permitted Telefonica to rely on an absence of inculpatory documents.

However, the criticisms made of the judge under this ground do not take full account of his reasoning. The judge was obviously well aware of what Telefonica had failed to do, including that specific preservation measures were not taken in respect of Mr Alierta until early 2020, and that there was no proper excuse or explanation (judgment at [57] and [303]). P4u also now accepts that the judge was correct to apply the approach in Efobi, so there was no error of principle. The judge considered the factual position in some detail and made strong criticisms of Telefonica’s approach. He was entitled to accept Mr Alierta’s own statement that he had not knowingly destroyed relevant documents, while recognising that relevant documents might nonetheless have been destroyed. Mr Draper submits that the judge was wrong to label such destruction as “innocent” and to reject P4u’s allegation of deliberate document destruction (judgment at [58] and [59]), but that imports too much close legal analysis into an area where common sense should prevail. Of course the judge knew that, by choosing not to put in place preservation measures, documents could be lost and so in that sense there was a deliberate decision, but the judge was answering a straightforward factual allegation of deliberate document destruction.

Further, the judge clearly had regard to the witness evidence of Ms Nicola Bridge, a partner at Telefonica’s solicitors Mishcon de Reya (judgment at [59]). In relation to Ms Castillo, that evidence confirms among other things that all her files, including emails, were backed up weekly and retained. As regards Mr Alierta, around 78,000 emails were retrieved, but it was also confirmed that he relied on his personal assistant for most electronic communication and did not have a smartphone.

More significantly in relation to Mr Alierta, P4u cannot escape from the strength of the judge’s findings that there was no collusion between him and Mr Colao, supported not only by Mr Colao’s witness evidence but by documentary evidence, including as to Vodafone’s then position: see [50] and [273] above.

Overall, I can see no error of approach by the judge in relation to Telefonica’s failure to adopt document preservation measures when it should have done. In particular, the judge considered whether to draw an adverse inference in relation to Mr Alierta’s position but decided not to do so. That decision was open to him.

As to the complaint that the judge should at least not have allowed Telefonica to rely on an absence of documents, Mr Draper referred to an instance of this at [133] and [138] of the judgment. This related to a discussion that Mr Dunne and Mr Evans of O2 had with Ms Castillo on 3 October 2012 in preparation for a wider meeting between O2 and Telefonica on 9 October. One of the points raised in a follow up exchange between Mr Evans and another colleague at O2, Mr Phil Maple, was entitled “Regional engagement and support to [sic] our move to exit an Indirect”. Mr Maple suggested “presumably the Region can guide us as to whether there is a broader appetite for change?” After considering the 9 October meeting the judge concluded at [138] that Mr Maple (who did not attend that meeting) may have hoped to gain some reassurance from Telefonica as to the likely market reaction to O2’s decision, but his comment did not bear the weight that P4u sought to place on it. The judge commented that there was no support in any subsequent internal documents for the suggestion that Telefonica had provided any information about the position of other MNOs at the 9 October meeting, and that the allegation was not put to Mr Dunne or Mr Evans.

Again, in my view the judge was entitled to refer to the absence of documentation. There was documentary evidence about the 9 October meeting, in particular an internal O2 email reporting “main takeaways” from it, which contains no hint of support for P4u’s allegation. If anything had been said by Telefonica then that might well have been expected to be picked up somewhere in O2’s own internal documents, given its importance to O2. Indeed, my reading of this part of the judgment is that it was O2’s documents, not Telefonica’s, that the judge likely had in mind.

I would therefore dismiss ground 7.

In conclusion, I would dismiss each ground of appeal. The judge made no material error of law and, in respect of challenges based on delay or compartmentalisation, I am satisfied that the judgment is safe.

In closing, I wish to say something further about delay in producing judgments. As already noted, in Bank St Petersburg the Chancellor referred at [78] and [84] to an unwritten rule that judgments should be delivered within three months of the hearing, and confirmed that this principle applied even in long and complex cases.

There are sound reasons for the existence of a principle of this kind. As the Chancellor said, justice delayed is justice denied, and significant delays lead to a loss of confidence in the justice system (see also Kanaya Dansingani v Canara Bank [2021] EWCA Civ 714 at [86], per Arnold LJ, citing Goose v Wilson Sandford & Co [1998] TLR 85 at [112]).

Nonetheless, three months cannot be an inviolable deadline in all cases. Whether a delay is excessive in a particular case will depend on a multiplicity of factors which will include not only the length of the delay but the length and nature of the hearing, the number and complexity of the issues, and the extent of the documentary and oral evidence to be grappled with. Further, while extrinsic matters such as the personal circumstances of the judge and the pressure of other work cannot detract from the effect of the delay on the parties to the case (albeit that they may well be mitigating factors in relation to the judge’s own position), the same does not apply to delay attributable to the need for the judge to take an appropriate level of care in producing his or her judgment. An inadequate, rushed, judgment may well also deny justice.

I have already noted, at [262] above, that the level of care taken by the judge in this case is obvious from his judgment and also from the fact that he re-read all the documents referred to during the course of the trial (in addition to many additional documents footnoted in closing submissions). This would have been an enormous task, but it can hardly be criticised given that the judge was deciding a complex, hard-fought case which involved levying a number of very serious allegations against senior executives of the defendant MNOs, allegations which, as the judge also observed in the permission judgment, would ordinarily be the subject of an investigation by a team of officials at a competition authority. And it was that level of care which has, in turn, enabled this court to conclude that the judge did not err.

I do not make these points in order to excuse a 15 month delay. Rather, I have set them out in order to emphasise that there is a wide variety of cases, and the extent to which a delay is prejudicial will depend on the circumstances.

This leads me to my next, and final, point, which is to reemphasise what I have said above (at [220] and [224] in particular) about the acute danger of island hopping and the court’s response to that risk. As Lewison LJ said in Fage v Chobani, the trial judge will have had regard to the “whole sea” of evidence presented to him, whereas an appellate court “will only be island hopping”. That risk is further increased when an appellate court is invited to apply an additional level of scrutiny to a delayed judgment.

It is critical that the additional level of scrutiny required of a delayed judgment is not permitted to become a means of enabling disappointed litigants to benefit from a focus on specific points that they wish to challenge, without a proper regard to the judge’s findings on all of the evidence. The court must therefore carefully consider the judgment, and the judge’s consideration of the other evidence, as a whole. And the parties should, of course, appropriately assist the court in that task, as indeed occurred in this case. Any other approach would not allow the true materiality (or otherwise) of the evidence on which the appellant relies, and thus the potential for injustice, to be determined.