Encountering complex scientific assessments in reviewing risk regulatory measures, the EU Courts habitually proclaim substantive deference and procedural strictness in their review alike. Application of this standard, though, appears as a case-dependent matter of degree and delineation. As the recent CWS Powder Coatings rulings and their scholarly discussions illustrate, the right degree of scrutiny may be a matter of perspective. Revisiting and contextualising this case, this post reflects on its implications for the judicial boundary between permissible and impermissible subjects of review. It submits that this doctrinal boundary is the product of recursive, judicial boundary-work. The question, then, is not only how far is far enough but also how the boundary is engendered and justified.
Legal doctrines are legal knowledge, i.e. not just knowledge about the law, but knowledge internal to the law. EU law’s doctrine of deference under circumstances of technical or scientific complexity is a particular specimen of its kind. It is legal knowledge reflecting its own limitations. Since early case law, the EU Courts recognise “broad discretion” where regulators are tasked to conduct complex assessments. Such administrative discretion implies judicial deference (Galligan, p. 5). Thus, substantive judicial review is commonly limited by the manifest error standard, barring the judges from substituting their appraisal of the facts for that of the regulators. Deference cannot be absolute however. What is at stake, then, is the delineation between what properly belongs to the justiciable and the juridical and what lies on the other side of that boundary, to wit: what is scientifically complex.
This boundary assumes particular significance where scientific knowledge is a well-institutionalised feature of the administrative architecture as is the case in EU risk regulation. From chemicals and pesticides to food and genetically modified organisms: regulatory decision-making organised by risk analysis as a procedural structure operates through a process seeing to scientific risk assessments informing and often guiding regulators’ decisions. Where risk regulatory measures become subject to judicial review, proceedings before the Courts often concern, at least ostensibly, less the infringement of specific rights than the sufficiency of their scientific substantiation. It is here that this doctrinal boundary may prove contested.
One of the most recent rows in the EU Courts’ case law struggling with this delineation is the CWS Powder Coatings case. Subject of the case was the Commission’s classification measure of Titanium Dioxide (TD) under the EU’s Classification, Labelling and Packaging Regulation for being potentially carcinogenic as a powder. Whereas the first-instance judgment had been discussed by Weimer and Morvillo, Bailleux has more recently examined the ECJ’s judgment on appeal delivered in 2025. Each discussion of the respective judgments offered criticism, yet pointing in seemingly opposite directions. Where Weimer and Morvillo argued the General Court had gone too far, Bailleux found the ECJ had not gone far enough.
Nevertheless, both positions may be understood to share a common methodological position. In search for this commonality, we will, first, briefly revisit the CWS Powder Coatings and, secondly, contextualise it in the broader stream of case law concerning the Courts’ deference under circumstances of scientific complexity.
Going too Far?
Although the case was multi-faceted, the focus may be limited to a single claim for the present purpose. The applicants argued before the General Court that the methodology used by European Chemical Agency’s (ECHA) Committee for Risk Assessment (RAC) to assess lung overload was inappropriate (Joined Cases T‑279/20 and T‑288/20 and Case T‑283/20, para. 80). The Heinrich study, which the regulators relied on, did not indicate the density of particles used in the study (ibid., para. 92). In applying the so-called Morrow calculation to that study, the RAC dealt with such uncertainty by applying a standard particle-density value of titanium dioxide (ibid., paras. 88 and 93). That standard density value, in the applicants’ view, did not properly account for TD particles’ tendency to agglomerate, which would affect lung overload potency. Excessive lung overload in the trials could invalidate the Heinrich study’s results.
Although expressly stating that determining the correct density value was not for it (ibid., para. 97), the General Court scrutinized the scientific assessment in considerable detail. It determined the relevance of both, particle density and agglomeration, through exegesis of the Heinrich study and the RAC Opinion (ibid., paras. 88 and 98). Against that background, the General Court held that the RAC had failed to take account of relevant factors by using the standard density value instead of nano-sized TD particles’ agglomerate density and, therefore, committed a manifest error of assessment (ibid., paras. 100 and 103).
As Morvillo and Weimer observed, the Court apparently sees itself “competent to identify the decisive elements in the classification (…) and to disagree with ECHA’s assessment”. In their view, the Court thereby exceeds checking for manifest error and blurs the line between what it promises to do (procedural review) and not to do (substantive review). AG Capeta took a similar position in her opinion on the appeal lodged against the General Court’s ruling: permissible review confines itself to annul a measure where a relevant fact was plainly ignored, however not for choosing one interpretation of such fact over the other (ibid., para. 91). As the General Court had substituted its appraisal of the correct density value for that of the regulators, it had decided what it should have deferred (ibid., para. 99).
Or, Not Going Far Enough?
The ECJ policed the boundary of permissible judicial scrutiny in its appeal ruling. In that the GC had substituted its own appraisal as to the appropriate particle density for performing the calculation for that of the RAC, it went beyond what the duty of care permits (Case C-71/23 P, para. 114 and 116). Applying the duty of care itself, though, it still found fault in regulators’ reasoning. That fault, in the ECJ’s view, was not an incorrect appraisal but a lack of engagement with the uncertainty surrounding particle density (ibid., paras. 118-124). Substituting its own reasoning for the General Court’s, the ECJ ultimately upheld the annulment (ibid., para. 125).
As argued by Bailleux, what the ECJ deems permissible and what it holds for impermissible review, however, is inextricably intertwined. The ECJ itself can only arrive at its conclusion for lacking consideration of particle density by presupposing this factor’s scientific relevance. As Bailleux (p. 661) notes, the Court’s distinction becomes “byzantine” – hiding behind nitty-gritty differentiations what should be done openly and is done anyways, i.e. more substantive scrutiny.
Ruptures and Shifts
The two rulings in CWS Powder Coatings may be a rare example of an explicit disagreement concerning the limitation of judicial scrutiny vis-à-vis complex scientific assessments between the two Courts. It is, however, only a recent example in a larger line of cases re-working the boundary between the permissible and the impermissible object of review. Within this lineage, the doctrine of deference’s figuration has undergone considerable change and gained in complexity.
This lineage points back to the 1980s. In rulings concerning the use of growth hormones in livestock farming (esp. here and here), which had been the subject of major political controversies in the 1980s, that delineation between permissible and impermissible objects of review is neatly drawn. The reach of each appears to have been settled through an apparent strict opposition between matters of law and those of science. Whilst the Court addressed the legal questions, its reasoning kept remarkably silent on issues of scientific controversy underpinning the measures at stake (Vos, pp. 146-7).
With increasing reliance on scientific knowledge by EU regulators, such super-deferential, “reductionist approach” had become “no longer acceptable” (Ladeur, 1478). By its 2002 Pfizer-ruling, it has also become obsolete. Here, the Court began justifying its doctrine of deference as a juxtaposition between effective judicial protection and institutional balance. Folding these competing ratios into a synthesis of ‘process-oriented review’, the General Court has pledged substantive deference yet procedural strictness (ibid., esp. para. 169-71).
Once coined modern, this approach has become regular in that it provides the basic tenets for subsequent case law. Yet, it has not informed a uniform practice (e.g. Fritzsche, 379; Morvillo and Weimer, 537; cf. also Knuth and Vos). Its doctrinal development often takes more subtle forms than the rupture represented by Pfizer. Subsequent rulings have refined many of its elements such as the precautionary principle’s scope, or introduced new variations into the rhetorical repertoire.
Hybrids
Even more so, under similar phrases quite divergent levels of scrutiny are enacted. In some instances procedurally-oriented review plays out more deferential (e.g. in GE Healthcare). Yet, the General Court’s ruling in CWS Powder Coatings (para. 40) is not a standalone example for its potency to delve deeply into scientific issues. Ironically, the ECJ itself rendered a procedural defect by virtue of its reasoning into a quasi-substantive one in its 2017 Bilbaína de Alquitranes ruling, without departing from recognizing broad discretion. The Courts’ reiterative invocations of substantive deference and procedural strictness appear as a Fishian (esp. p. 199) narrative dichotomy: standards remain uniform, application varies. Where the duty of care serves as a tool to penetrate into the scientific substance, eventually, its application renders the distinction between substance and process porous.
Such hybridity arguably is not merely a result but also a feature of the primary doctrinal vessel for such tightened scrutiny. Often enmeshed with the manifest error-standard as had been done in the CWS Powder Coatings rulings (Nehl, p. 193), the duty of care serves judges not only as a quantitative check for completeness in reasoning, but also as a qualitative standard for assessing the probative value of evidence (Hofmann, 98-102). Yet, relevance is an empty conceptual shell. Nothing is relevant in itself. It takes relation to generate relevance through value judgements (cf. Nehl, p. 177). Where legal norms lack specification, judges may turn towards scientific heuristics. This, too, is not a feature of the General Court’s assessment in CWS Powder Coatings alone. Assessing the reliability of a toxicological study, as the General Court has done in its Sasol Germany GmbH and Others (paras. 73 et seq.) ruling, the duty of care may serve judges as a tool to verify whether that study was correctly classified under the established Klimisch scoring system. The duty’s hybridity is hence not only one between procedure and substance, but also one residing in its potential to integrate extra-legal relevance-criteria into judicial reasoning.
Drawing Lines in the Sand
Effective judicial protection necessitating rigid control of the exercise of public authority is a value that must appear as self-justifying for a lawyer. Yet, there is also good reason to keep judges from strolling into scientists’ domains. Themselves institutions in-between institutions, the Courts, too, have to respect the institutional balance even if they are the principle’s authoritative interpreters. As noted in AG Capeta’s CWS Powder Coatings opinion (para. 86): “The courts are not scientists and cannot become scientists”.
There are good reasons pointing either way. Yet, arguably none of those fully determines the form deference should take. In contrast to the Courts’ occasional ontologizing reference to the ‘nature’ of scientific knowledge, the boundary may not be determined through science’s otherness as an external referent (Fritzsche, 380). Neither science nor law have footholds outside the ever-changing spheres of human culture. Similarly, juxtaposing effective judicial protection and institutional balance may provide guideposts but achieving that balance remains an interpretative effort lacking finality under varying contextual parameters. The “delicate balance” is not simply there, it has to be struck.
The doctrine of deference’s continuous variation and the ambiguity of those distinctions by which it is enacted, may allow for another conclusion than criticism as too much or not enough of scrutiny. In that, both positions criticizing the CWS Powder Coatings rulings can be read as converging. Reiterations of identical phrases promising restraint where substantive scrutiny is exercised threaten to become façade. So, too, do increasingly ‘byzantine distinctions’.
Perhaps the doctrinal boundary does not merely represent science as law’s natural “other”. Drawing it arguably contributes to producing that distinction whilst simultaneously presenting it as if it had been there all along. What Thomas Gieryn observed in scientists’ rhetorical construction of disciplinary identities may thus be transposed to the judiciary’s deference towards scientific complexity: the boundary is not an external barrier but a product of judicial “boundary-work”. This, arguably, is what can be observed, too, in CWS Powder Coatings. Even where demarcations shift, new boundaries are drawn with even greater rigour.
Even if lines can only be drawn in the sand, these lines may still be drawn differently. With scientific and legal knowledge not being separate territories but having to be separated time and again, their relation is itself as ambiguous as those lines that intend to separate them. Law’s encounter with scientific knowledge in the courtroom thus points back at the judicial raison d’être: doctrinal reasoning. Where things are determinate, reasoning becomes decoration. Where things are ambiguous, reason-giving may not resolve with finality but at least structure. Being subject to authority that has to justify itself, arguably, rests at the core of effective judicial protection. This holds, too, for the principle’s primary agents when enacting deference where matters become complex. The doctrine of deference may not resolve the ambiguity in the relation between what it seeks to separate. Yet, it must provide reasons for the operations by which that boundary is drawn.
Posted by Luca Knuth (luca.knuth@maastrichtuniversity.nl), PhD researcher at Maastricht University and Kiel University.
This post is part of a thematic section on REALaw blog on the epistemic boundaries of EU courts, which aims to bring together contributions that analyse how EU courts access, interpret, and evaluate technical and factual information. Together, these contributions assess whether the resulting epistemic boundaries remain adequate in light of contemporary expectations of effective judicial review. This section is coordinated by Dr. Pietro Mattioli, postdoctoral researcher at the University of Maastricht. His research deals with the growing technical specialisation of administrative decision-making and its implications for judicial review.