CA-2024-001499 - [2025] EWCA Civ 1321

I am acutely aware of the need for this court to respect the function of the trial judge, and show due restraint. Having reflected on the arguments and revisited the written materials presented to us I have however concluded that the judge’s approach was in some respects wrong in law in ways that are material to the outcome.

I would reject ground of appeal 1(a). It was open to the judge to conclude that the reputational impact of the claimants’ tweets was mitigated to some extent by the fact that they were manifestly expressions of opinion. In my judgment, however, Mr Fox’s appeal should be allowed on grounds 1(b) and (c). The judge inferred from third-party publications and specific incidents that Mr Fox had acquired a bad reputation for being a racist. That was contrary to long-established common law principles, which apply equally when assessing serious harm. In addition it was in part unsound for procedural and evidential reasons. I would also allow the appeal on ground 2. In my judgment the evidence compelled a finding that each of the claimants’ tweets caused serious harm to Mr Fox’s general reputation. The judge’s findings that the tweets did not cause harm to his agency relationship or career are vitiated by legal error: the judge applied, in material parts, the wrong legal test on causation. I would therefore set aside the judge’s order dismissing Mr Fox’s claim and remit the claim to the High Court for retrial on the remaining issues. My detailed reasons follow.

The judge’s reasoning on this issue is elaborate and sophisticated. Not all of it is entirely easy to follow at first sight. I confess that the relevance of some of the points made (such as Mr Fox’s standpoint on the right to express views on what is racist) continues to elude me. But at the core of the relevant passages are, I think, two main points.

The first is that an expression of opinion may, depending on the context, have a less harmful impact on a person’s reputation than a simple statement of fact to the same effect. As a general point of principle I do not think this is open to sensible dispute. True, as Nicklin J pointed out in Morgan (above) at [18], an expression of opinion can cause reputational harm, or there would be no need for the defence of honest opinion. That is axiomatic. But it does not follow that comments are necessarily as reputationally damaging as assertions of fact. This is a point implicit in the reasoning of the Court of Appeal in Singh v British Chiropractic Association [2010] EWCA Civ 350, [2011] 1 WLR 133, in particular at [32]. In Sube v News Group Newspapers Ltd [2018] EWHC 1234 (QB) I reflected further on the issue, concluding that on the particular facts of that case the expressions of opinion complained of did not have a tendency to cause serious reputational harm. In Morgan (above) Nicklin J reviewed these cases and at [31] extracted some general principles as to the court’s approach to serious harm in opinion cases, including these:

(iii)

the significance of an imputation and its capacity (or tendency) to cause serious harm to reputation may be affected by its context and presentation …;

(iv)

the fact that the opinion is clearly presented to the reader as such may well mitigate its defamatory impact;

(v)

if the source of the criticism is identified, does s/he appear authoritative? Is his/her view likely to carry weight and be accepted by the reader? Or is the critic someone whose view the reader is likely to discount in favour of making his/her own assessment?

The language here, as in Sube, reflects the interpretation of s 1(1) adopted by the Court of Appeal in Lachaux, which was authoritative at the time. Point (iii) should now be read as referring to the likelihood that the imputation will cause serious harm rather than its tendency to do so. With that qualification I would endorse this summary.

The second main point implicit in the judge’s reasoning is that the present case is one of those in which the nature of the statements, their context, and their sources did mitigate the harmful impact of what was said. Her thinking was, as I read it, that Mr Fox’s Sainsbury’s tweet was part of the context in which the claimants’ tweets were published such that many readers will have seen their use of the damaging term “racist” as (a) an expression of opinion by a celebrity using social media to take a position on a topic of current interest, prompted if not provoked by some outspoken remarks of another public figure in the form of Mr Fox, as opposed to, for instance, (b) an authoritative and considered finding of fact arrived at by an academic or other professional on the basis of months of diligent and rigorous research. This analysis explains, in my view, why the judge deemed it relevant, in this connection, to mention Mr Fox’s political stance, that he had stimulated a debate about racism, and that “some might think” he had done so provocatively. It also explains the judge’s reference to what the tweets might tell readers about the claimants.

Some criticisms can be made of the judge’s reasoning. Among them is that, unlike Mr Blake and Mr Seymour, Ms Thorp did not indicate to her readers the basis for the opinion she expressed. Looked at in the round, however, my judgement is that the judge’s approach was in conformity with principle and her assessment was one she was entitled to make. This aspect of the judgment reveals no error that satisfies the Re Sprintroom test. I do not think the same can be said of the judge’s approach to the reputational impact of third-party publications and Mr Fox’s own conduct. It shall take these points together, in that order.

The relevant passages in the judgment are in the section on the “inherent probabilities and the balance of the evidence” (the judge’s factor (d)). I have summarised this at [54]-[58] above. The key features for present purposes are these. Having disposed of the issue of whether Ms Latimer dropped Mr Fox because of the tweets complained of, the judge went on to deal with the two other limbs of his case on serious harm: “Mr Fox’s career and reputation more generally”. She addressed those two limbs of the case together, compendiously, without drawing any clear distinction between them. The judge began her analysis of this part of the case by identifying it as one involving “issues of causation”. It was in that context that she made a series of points about Mr Fox’s career path, “views” that other people had taken about him, his admissions about such views, things that other people had said about him, and things that he had said. There was nothing wrong in principle with reliance on admissions by Mr Fox. It is well-established that these are admissible. But the judge’s approach was flawed in other respects.

It was a mistake to conflate these two strands of Mr Fox’s case. His contention that the claimants’ tweets had destroyed or damaged his career had a connection with his claim that he had suffered general reputational harm; it was relied on to illustrate and support that claim. But it was a separate, distinct and narrower strand of his case, and it raised different issues which needed to be analysed separately.

I shall deal with the Dingle issue first. There was evidence before the judge that other people had referred to Mr Fox as a racist before the tweets complained of. It was, indeed, his own case that some people had done this. He pleaded it, and he admitted it in his evidence at trial. But it was emphatically not Mr Fox’s case that he had a general reputation as a racist before that allegation was made against him by the claimants on 4 October 2020. On the contrary, Mr Fox maintained that he acquired a general bad reputation of that kind after and as a result of the claimants’ tweets. In their defence to his claims, the claimants partly admitted Mr Fox’s case that there had been some third-party allegations against him before the tweets. It was in that context that they identified the 15 tweets. That was, in substance, all that was put to Mr Fox in cross-examination. Mr Fox complained at the time that this part of the claimants’ case contravened the rule in Dingle. I do not think it did. The claimants did not plead the 15 tweets as evidence of a pre-publication general bad reputation as a racist, nor was that the way they put the case at trial. That would clearly have contravened the rule in Dingle. The claimants did plead that Mr Fox had such a general bad reputation. But they did that separately and without reference to the 15 tweets or any other specifics. Presumably, it was at that time the claimants’ intention to prove bad reputation in the manner permitted by the authorities: by calling witnesses who could say from their own knowledge that Mr Fox had a settled general reputation of that kind. But they served no statement from any such witness. And by the time of trial they had abandoned that allegation. No issue of existing bad reputation was before the judge for adjudication.

The judge’s reasoning assumes that a distinction is to be drawn between reliance on third-party publications as evidence of an existing bad reputation and reliance on the same material as going to the issue of causation of general reputational harm. This judge has made similar observations in other first-instance decisions: see for instance Sivananthan v Vasikaran [2022] EWHC 2938 (KB), [2023] EMLR 7. In Peros (above) Applegarth J reviewed the Australian and English authorities, including Sivananthan and the first instance judgment in this case, and concluded that this distinction was legitimate and consistent with Dingle. He said (at [245]) that there was no reason in principle why a defendant contesting an allegation of serious harm should not be able to tender third-party publications “and submit that in the circumstances they affected the reputation of the claimant among readers, including readers of the subject publication”. Applegarth J recognised that this was a subtle distinction but considered it a valid one. He observed (at [246]) that it is “no different in principle from an injured worker, who claims that his bad back was caused by a certain incident, being asked about other incidents that may have caused that injury, and the defendant calling evidence about those other incidents.”

I think this is a fair description of how Collins Rice J approached the present case. But I think it is wrong. Mr Fox sought to show that on and after 4 October 2020 he had a general bad reputation as a racist which was caused by the tweets complained of. One strand of his case relied on specific events to illustrate the reputational harm alleged. To that extent the analogy identified by Applegarth J is sound and reliance on third-party publications is unobjectionable. But Mr Fox also, separately, invited the court to infer that the widespread publication of a seriously damaging allegation caused serious harm to his reputation generally. In this context, the analogy with a claim for personal injury is not apt. In evidential terms, an injured reputation is not like a broken leg. It consists of the esteem in which a person is held by others. As I commented in Wright v McCormack (CA) at [50]-[52], that “is not so easily observed or demonstrated” as a physical condition; but the law presumes that claimant’s reputation before publication was good; it is rare that the claimant can adduce affirmative evidence on the issue of reputational harm; and “the assessment of harm is often a matter of inference from …. the gravity of the imputation and the nature and extent of publication …”.

When it came to this part of the case the first step for the judge, logically, was to assess whether the publication of the offending statements had probably caused some reputational harm that was serious. This does not have to be a complex exercise. As the judge observed at [64], when addressing the claimants’ case on serious harm, “this is a threshold test ... not an exercise in definitively quantifying harm caused”. I would agree with the way the claimants put it in their trial skeleton argument, when laying out their own case on serious harm: “In any case as to mass publication, there is a likely prospect that some individuals who read the publication will have thought less of the claimant in consequence, whereas others will have been indifferent or otherwise as to the publication. Serious harm is not a ‘numbers game’ … reputational harm in the eyes of some of the readership is sufficient - the issue is one of threshold liability: thereafter the extent of harm becomes relevant in terms of damage.”

The judge took a different approach. She made no finding as to the state of Mr Fox’s post-publication reputation generally. Instead she asked herself whether, if Mr Fox had a reputation for being a racist, that reputation had been shown to result from the tweets of which he complained. She then identified and considered a range of other factors that might have caused or contributed to reputational harm of that kind and concluded that Mr Fox had not proved that any  harm was not caused by these other factors (or not to a degree that was serious).Among the factors the judge relied on were the inferred reputational consequences of previous third-party publications to the same effect as the alleged libels. I find this an unsatisfactory reasoning process. And I am unable to see how it can be reconciled with the authorities. The proposition that s 1 of the 2013 Act authorises reliance on third-party publications to assess the causation of general reputational harm, assuming some was caused, seems to me to be no more than a recast version of the argument rejected at all levels in Lachaux.

I do not believe the process undertaken by the judge can be justified as an exercise in “isolating” the damage caused by the alleged libels. That term is to be found in various passages in Dingle, most famously in these words of Lord Denning:

If the judge isolated the damage for which the ‘Daily Mail’ were responsible from the damage for which they were not responsible, he would have been quite right, see Harrison v Pearce [(1858) 1 F & F 567]. But it is said that he did not isolate the damage. He reduced the damages because the plaintiff’s reputation had already been tarnished by reason of the publication of the report of the select committee and of the privileged extracts of it in the ‘Daily Mail’ and other newspapers. I think he did do this and I think he was wrong in doing so.

Over the years the notion of isolation has proved somewhat elusive. The speeches in Dingle indicate that it denotes a process of ensuring that a claimant is not compensated for reputational harm caused by others. For that purpose it may be legitimate to have regard to the fact that there were other publications to the same effect. The more recent cases I have mentioned at [30] above treat “isolation” as including reliance on third-party publications to rebut a claim that specific consequences have flowed from the publication complained of. That is surely correct. There may be more to isolation than this. But one thing is clear: the House of Lords were not using the concept of “isolation” to embrace a process of identifying earlier third-party publications as candidate causes of reputational harm, inferring that they did indeed operate in that way, and moving from that finding to the conclusion that the claimant had an existing bad reputation so that the publication complained of was less reputationally harmful. That is not something covered by the notion of “isolation”. It is, in my opinion, a line of reasoning unequivocally prohibited by Dingle, Lachaux and Mueen-Uddin. Yet it is one that the judge adopted here. I can see nothing in the claimants’ suggestion that Mr Fox somehow waived his right to rely on this rule.

I find the judge’s approach to the evidence of the 15 tweets problematic, quite apart from Dingle. In a case of limited publication, or one where the claimant points to some specific event as probative of reputational harm, evidence of this kind might be important. In such a case statements by the publishees can go to show that little or no harm was done: cf Amersi v Leslie (above) at [197]-[219]. But that is not how the judge put the matter here. She was right not to do so. Reasoning of that kind cannot work in a case like this, where a few tweets are relied on in the context of a complaint of publication to hundreds of thousands of individuals. The inescapable conclusion is that the judge treated the 15 tweets as a basis for inferring that before October 2020 Mr Fox had acquired a bad reputation among those to whom the claimants’ tweets were published. That, however, presupposes at a minimum that the 15 tweets or some of them were widely published and read by people who later read the tweets complained of. Nothing was pleaded on that score, and the judgment contains no analysis of the scale of publication of the 15 tweets or the extent to which their readership overlapped with that of the tweets complained of. The same goes for the judge’s broader references to what people had made of Mr Fox’s Question Time appearance.

The judge’s reliance on things previously said by Mr Fox himself raises different issues. The statements referred to by the judge were relied on by the defendants as previous (racist) misconduct of Mr Fox. As such, they were subject to the exclusionary rules mentioned at paragraphs [28(1) and (4)] above. They were properly in evidence at the trial nonetheless because they were (a) matters relied on by Ms Thorp in support of her defence of truth and (b) matters which all three defendants said should operate in mitigation of damages pursuant to Burstein or Pamplin. But the judge never got to the issue of truth, or the question of damages. She considered these matters only in connection with the issue of serious harm. In doing so, the judge did not rely on the Burstein principle (and it has not been suggested that she could or should have done so on the facts). Nor did she refer to or rely on Pamplin (and she would have been wrong to do that, for the reasons explained in Wright v McCormack (CA)).

Instead, the judge took an approach similar to the one she followed in respect of third-party publications. In substance, she accepted the argument for the claimants. Boiled down to its essence, that argument was that Mr Fox’s case on serious harm should be rejected because he had made some statements that were controversial; and even if these were not instances of racist behaviour, as alleged by Ms Thorp, a lot of people will have thought they were, and it is this that will have caused any reputational harm. This summary may be crude, but I think it is fair. In my judgement this approach is not only unattractive but wrong in principle.

There are several difficulties with it. The first is that it adopts the same unsatisfactory approach of assuming or hypothesising harm as a prelude to examining causation. The second problem is that the submission fails to observe the well-established distinction between the approach to allegations of general harm and specific instances. It is, as I have said, clear law that a defendant may point to alternative factors as explaining specific harm or a specific event which the claimant says was caused by the alleged libel. It is however impermissible to rely on specific acts of the claimant to show that he already had a general bad reputation at the time the alleged libel was published. This is not something prohibited by the rule in Dingle, which is about other publications to the same effect as the alleged libel. The claimants were right about that. But the approach cannot stand with the earlier decision of the House of Lords in Plato Films Ltd v Speidel (above).

The plaintiff in that case was the former Joint Allied Commander, Europe. He sued for libel in respect of some specific accusations of war crimes. The issue before the Lords was whether the defendants could plead in mitigation of damages (A) that the plaintiff had a general bad reputation for having committed those crimes and/or (B) that he had committed other specific wrongs of a similar kind. Affirming the decision of the Court of Appeal on these points the House held that plea (A) was permissible, but that only general evidence of reputation could be given to support it; plea (B) was impermissible, as evidence of specific acts may not be called to prove that the plaintiff is of bad character or disposition. The House considered these points to have been established by Scott v Sampson (1882, above). I do not think it was the claimants’ case on serious harm that Mr Fox had a “bad character” as a racist or that the judge made a finding to that effect. The issue they addressed was his reputation. But in my opinion the claimants’ argument and the judge’s decision on that issue were contrary to the reasoning of the majority in Plato in respect of plea (A). Lord Radcliffe thought that the “general evidence” of bad reputation which a defendant could call might include specific instances, provided they were sufficiently notorious. But he was alone in taking that view. The established law was summarised by Lord Denning at p.1140, having reviewed the authorities: “When general evidence of bad character is given, the witness cannot in chief give particular instances”.

This rule has stood for well over a century. Its validity has not been challenged on this appeal. Nor has it been suggested that it is inapplicable when considering the issue of serious harm. I think it must apply. I see nothing in the language of s 1 to cast doubt on that. Nor am I aware of anything in the legislative history that might do so. And the point Lord Sumption made in Lachaux applies equally here: it would be irrational to apply such a rule in the context of damages but ignore it when addressing the related issue of serious harm. Yet it was not addressed in the judgment. It does not appear to have been confronted by the claimants at the trial.

I think there is an additional problem here. The judge’s discussion of the causative role of “that which is called out” ([57] above) involved an implicit finding that Mr Fox’s Sainsbury’s tweet led readers other than the claimants to think he was a racist. It was however a central part of Mr Fox’s reply to the honest opinion defence that this was a reasoning process that no honest person could follow, and the judge did not address that issue. I reach no conclusion on this specific point, on which we did not hear argument. The other flaws in the judge’s reasoning are enough to vitiate her conclusion that Mr Fox’s case on serious harm should be rejected.

In my opinion, the only conclusion reasonably available to the judge was that the publication of each of the claimants’ tweets did cause serious harm to Mr Fox’s reputation generally.

I do not think we are concerned here with a single indivisible head of injury, as suggested on behalf of Mr Fox. Each claimant published separately from each of the others. Each had their own group of readers. It may be that there was a degree of overlap between the groups, but the judge made no finding that there was. I have not detected that there was any evidence on that issue. In these circumstances, each claimant is responsible for the harm caused by the publication of their tweet to the readers of that tweet, but none can be held responsible for any harm caused by another’s publication of a different tweet to other readers. Put another way, each claimant is entitled to have the harm caused by their tweet assessed in “isolation”.

That said, each claimant was on the judge’s findings responsible for the “mass publication” of the allegation that Mr Fox is a racist. That allegation had a defamatory tendency. The judge held it was a serious allegation, though its impact was reduced to some degree by its status as an expression of opinion which Mr Fox had to some extent provoked. Mr Fox was presumed to have had a good reputation at the time of publication. He had made some admissions against interest about his pre-publication reputation, but these were plainly not admissions of a settled general bad reputation as a racist. The claimants did not pursue their pleaded allegation that Mr Fox already had a general bad reputation as a racist and there was no basis for a finding that he did. Applying the approach identified by the claimants ([78] above) these considerations provide compelling support for an inference that the publication of each tweet harmed Mr Fox’s reputation to an extent that is serious.

The remaining mitigating factor identified by the judge is the contribution which Mr Fox’s replies – the paedophile tweets - made to the circulation of the claimants’ defamatory statements about him. I do not see how this can materially undermine Mr Fox’s case on this issue.

For these reasons, I consider that the serious harm requirement was plainly satisfied in each case. It follows that the appeal should be allowed on ground 2. The case should be remitted for re-trial on the issues of honest opinion, truth and, if it arises, damages. I shall, however, address Mr Fox’s challenge to the judge’s conclusions in respect of his claim that the tweets were causative of Ms Latimer’s decision to end her agency relationship with him and the effective end of his acting career.

Those are two specific harmful events that were admitted or proved by direct evidence to have taken place after 4 October 2020. The issue in each respect was causation. The judge considered Mr Fox’s case that the claimants’ tweets caused each of these heads of harm and addressed a range of possible alternative explanations. After careful examination of the evidence she reached the conclusions I have quoted above and rejected Mr Fox’s case. I have looked closely at the language she used when explaining those conclusions. It is fair to say that her terminology was variable and inconsistent. At some points she considered whether the tweets complained of were “causative” of or had made a “material contribution” to the harm complained of. In these passages she was citing the appropriate test. But elsewhere she used different language, asking herself whether the claimants’ allegations were “themselves” causative of harm, or whether the harm was caused by those allegations “as distinct from” or “rather than” some other factor (such as other people’s allegations of racism, Mr Fox’s “chosen and sustained presentation” of himself as someone challenging public opinion on the issue, or other aspects of his own conduct). Some of this language suggests a test of whether the claimants’ tweets were the dominant cause of the harm in question. Some of it supports the interpretation adopted by the claimants: the judge was asking herself whether Mr Fox had shown that the claimants’ tweets were the sole and exclusive cause of the harm.

I have reminded myself that we should focus on the substance of a judgment and should not be derailed by purely formal or semantic criticisms. But I am satisfied this is a matter of substance not form. The judge’s conclusions on causation are flawed because they did not turn, as they should have done, on whether the offending statements made a material contribution to the specific heads of harm under consideration. That was an error of law.

I am not persuaded by the claimants’ objection that Mr Fox was not entitled to take this point on this appeal because he did not raise it at trial. The proposition of law is well-established. It was relied on in Mr Fox’s Reply which asserted that each tweet complained of was “a significant causative factor” in the harm alleged. It was clearly identified in Mr Fox’s trial skeleton argument as the test for identifying the reputational harm caused by the claimants’ tweets. The liability judgment stated (at [57]) that “the basic tort rules of causation” applied. And the claimants’ own trial skeleton argument asserted that “the court must consider … whether the publication materially contributed to harm”. In these circumstances I do not consider it is unfair to the claimants for us to consider and act on this ground of appeal.

I am not persuaded, either, that on a proper application of the law to the facts Mr Fox’s case on these points was bound to fail, as argued by the claimants. Nor am I convinced that it was bound to succeed, as Mr Fox contends. The error was plainly material. I would therefore uphold ground 2 for these further reasons and remit these issues for fresh consideration at the re-trial.