Experimental Evidence In Patent Disputes, Part 1: Lessons From UK Case Law

Experimental evidence can shape the outcome of patent disputes, but recent UK case law underlines the need for careful planning, procedural compliance and proportionate disclosure

Experimental evidence continues to play a pivotal role in patent disputes. It can be critical to the success of arguments on infringement and validity – but only when deployed carefully. The UK courts have recently addressed the scope of the disclosure that must accompany any experimental data, in Merck Sharp & Dohme v Halozyme.

We will examine these decisions and other key UK cases which offer valuable guidance on how litigants should approach the design, admission and evidential value of experiments in part 1 of this article. We will consider the approach taken at the Unified Patent Court (UPC) in part 2 of this article.

1. Introducing experimental evidence into UK proceedings

Courts in the UK recognise the potential value of experimental data, but there are strict rules on when and how it can be introduced. A party may seek to conduct new experiments during UK proceedings or to introduce post-filed data from non-UK litigation (e.g. from EPO oppositions). In either case, parties must obtain permission and serve a compliant Notice of Experiments under CPR PD 63A, which details inter alia the facts which the party seeks to establish and particulars of the proposed experiments.

Failure to follow procedure risks judicial criticism

In Eli Lilly v Genentech ([2018] EWHC 3522), the High Court criticised Lilly’s attempt to rely upon certain experiments conducted by Novartis for the purpose of EPO proceedings without permission. In seeking to rely on Novartis’ experiments without making a court application, the court considered that Lilly had proceeded with “brazen disregard of the Court’s orders and the well-established practice for admission of experimental evidence”. This ruling sends a clear message: evidence generated for EPO proceedings cannot simply be repurposed into UK litigation without engaging with the rules.

This was reiterated in Gilead v NuCana ([2023] EWHC 611 & [2022] EWHC 3042). Although the court allowed the experiments that NuCana sought to rely on from EPO proceedings into the case, it noted that simply including experimental evidence in a party’s expert report did not comply with the court’s rules. Further, a party seeking to rely on EPO-derived data may be ordered to repeat the experiment, but as Mr Justice Meade noted, such repetition is not always valuable: “sometimes in patent litigation experiments, the value of a witnessed repeat or reply experiment is actually minimal and many repeats and replies happen with very little value being extracted from them”. Gilead also cautions against relying on experiments that have been heavily influenced by lawyers or ‘super experts’, as this risks hindsight.

Practical takeaway

Strict procedural discipline is essential. If experiments were originally designed for EPO proceedings, think early about whether they remain fit for purpose in UK litigation and whether repeating them would help or hinder your case.

2. How UK courts assess the weight of experimental evidence

UK judges frequently emphasise that admissibility is not the same as persuasiveness. Recent decisions illustrate when experiments illuminate a technical issue – and when they fail to move the needle.

When a patent’s disclosure limits the value of experiments

The Court of Appeal recently handed down its decision in Salts v Pelican [2026] EWCA Civ 93 (19 February 2026), upholding the finding of non-infringement at first instance ([2025] EWHC 497). The first instance decision in this case serves as a cautionary tale as to the limited value of experimental evidence where there is little data in the patent.

Salts had alleged that Pelican’s ostomy bag infringed its patent, which addresses problems with unsightly bulging of ostomy bags. Both parties relied on extensive experimental evidence with respect to the issue of infringement. However, the High Court noted that the patent contained no measurements or experimental examples against which an alleged infringement could be assessed. Salts’ experiments, which consisted of filling the ostomy bags with various materials, were considered to be inconclusive, particularly given that both parties’ experts agreed that some approaches (such as taking plaster casts of the bags) were not industry standard practice. The court concluded that, whilst the experiments were helpful to demonstrate the behaviour of the ostomy bags in practice, none of them were decisive on the issue of infringement. It was difficult to draw any conclusions on the significance of the experiments, as the patent gives no guidance on what is significant in achieving the invention disclosed.

Practical takeaway

A patent that provides no framework for evaluation may reduce the evidential weight of even well-conducted experiments.

Statistical analysis: useful but not essential

While statistical analysis of experimental data may be helpful, lack of such analysis does not necessarily limit the data’s significance. In BioNTech v CureVac [2024] EWHC 2538, the parties disputed the validity of CureVac’s patents relating to a split poly(A) tail in mRNA said to ‘improve’ protein expression. BioNTech relied upon in vitro and in vivo expression studies to show that not all mRNA constructs tested with a split poly(A) feature (within the patent claims) result in improved protein expression. The court held that the patents were invalid for insufficiency and obviousness, finding that the claimed technical effect was not plausible, and was not in fact achieved, across the scope of the claims.

In his judgment, Mr Justice Meade addressed criticisms regarding the statistical significance of the experimental evidence. While acknowledging that statistical analysis has its use, he cautioned against over-reliance, noting that:

• Statistical analysis helps but is not required for data to have evidential value;
• Visual, qualitative interpretation of data could be considered as part of the evidential picture. Scientific plausibility was seen as a holistic assessment, not a purely statistical one;
• Criticisms of statistical methods affected weight, not admissibility; and
• Criticisms of experimental design will carry less weight when a party had the opportunity to comment on the protocol in advance but kept back its criticisms until it was too late to fix them.

Practical takeaway

Courts take a pragmatic, real-world approach to scientific data. Statistics support the analysis, but they are not a strict gateway to evidential value.

3. Impact of experiments on disclosure obligations

Experimental data often prompts disputes about disclosure – especially in pharmaceutical cases where lab notebooks, protocols and samples may be highly valuable.

Work-up experiments disclosable

Disclosure obligations may arise in the context of both litigation experiments and post-filed data from non-UK litigation introduced into UK proceedings. Where litigation experiments are conducted for a UK case, the courts have held that any work-up experiments are disclosable, and the service of a Notice of Experiments waives any privilege in those work-up experiments (this is referred to as Mayne Pharma disclosure as the requirements originated in Mayne Pharma Pty Ltd v Debiopharm SA [2006] EWHC 164 (Pat)).

More recently in Merck Sharp & Dohme (MSD) v Halozyme [2026] EWHC 1287 (21 May 2026), MSD was awarded indemnity costs after Halozyme brought a counterclaim for infringement of its patent relating to a modified hyaluronidase, PH20, stating that its belief that the patent was infringed was based on experiments conducted in parallel German proceedings. However, Halozyme did not provide either full disclosure of the German experiments, including work-up experiments, or the results of new experiments that it had sought to conduct for the UK proceedings.

Ultimately, Halozyme consented to revocation of its patent and MSD alleged that “Halozyme’s case on infringement was always either speculative, weak or thin”. His Honour Judge Hacon concluded that Halozyme could not have it both ways: having chosen not to provide full disclosure of either the English experiments or the German experiments, it could not also ask the court to assume that the results obtained supported its claim to having had a bona fide belief in its case on infringement.

A subseqent ecision in this case ([2026] EWHC 1838 (20 July 2026)) highlights the limits of the requirement for Mayne Pharma disclosure. The Patents Court considered the scope of the Mayne Pharma disclosure that Halozyme had provided alongside a Notice of Experiments relating to a second patent in dispute. MSD applied to the Court for disclosure of further documents (including unredacted copies of certain documents) and the dispute came down to the correct characterisation of the ‘issue’ in respect of which privilege had been waived by service of the Notice of Experiments. The Court ultimately concluded that what Halozyme had released from privilege represented the whole material relevant to the issue in question (which should be narrowly defined by reference to the Notice of Experiments), and the redacted material was irrelevant and privileged.

Disclosure must relate to pleaded issues

Where a party requests additional disclosure in light of pre-existing experimental data from other proceedings, the court may limit any further disclosure to what is necessary to assist the court with resolving the issues in the case.

In Akebia Therapeutics v FibroGen ([2020] EWHC 866 & [2019] EWHC 1943), the parties agreed on certain aspects of the scope of FibroGen’s disclosure, but Akebia sought further disclosure relating to experiments that it argued addressed the issue of insufficiency. The court considered that there was no justification for Akebia’s broad request as the additional documents were not relevant to the insufficiency case as pleaded. Further, requiring FibroGen to conduct searches for the documents would not be of any real assistance.

Access to samples of the alleged infringing product

In infringement cases, a party may want to conduct experiments on the alleged infringing products, but samples of the product may not be readily available. Helpfully, the UK courts have shown a willingness to order disclosure of samples for testing, for example:

In Boehringer Ingelheim v Generics (UK) Ltd (t/a Mylan) [2019] EWHC 584, Boehringer sought samples of Mylan’s alleged infringing products. In summary, granting the disclosure of samples, the court concluded that:

• The experiments should be conducted on the product samples in the form that would be used by the patient.
• A party seeking to resist disclosure of alleged infringing products must put forward its reasons for doing so, and the court would consider difficulties with obtaining the requested samples.
• Where relevant, the court will consider the principle of proportionality in making orders to disclose samples (here, the court considered that Boehringer’s request for ten samples was excessive and ordered disclosure of only five samples); and
• Requests to use samples in foreign proceedings may be permitted, subject to a separate application (here, the court left open the possibility of Boehringer making such an application in the future).

Practical takeaway

In appropriate circumstances, the UK courts are willing to order disclosure of samples for testing. However, parties should only ask for, or be required to provide, targeted and proportionate disclosure. Requests must be tied to the pleaded case and justified by the issues raised by the experiments.

Conclusions and key recommendations for litigants

• Plan early and respect formalities – experimental evidence can be powerful, but only when designed with the litigation issues and jurisdictional rules for admitting experimental evidence firmly in mind.
• Align with the patent – the UK courts will downplay experiments that lack a meaningful anchor in the patent’s disclosure.
• Prepare for disclosure requests – requests for disclosure or samples will be assessed based on necessity and relevance. Work-up experiments are likely to be disclosable.