If the scientist denies substantial forms in nature, he must still face the question, "Does science have a specific nature?" If it does not, then there is no better or worse––no more or less substantially actual––way to do science. Substance is an irreducible set (or mode) of essential features (or conditions) for a thing's existence. If science has no substantial form––on the supposition that there are no substantial forms as such––then there are no essential features for actually instantiating science. As such, a (version of the) scientific method inclusive of substantial forms (qua explanatory structures) just as well counts as science as a non-substantial version of the scientific method. In that case, though, the scientist has no scientific grounds for denying substantial forms in science. He just happens to prefer a non-substantial 'flavor' of science. To say, "I know science, and that is not it," is just a de-platonized way of saying, "I apprehend the pure Form of Science and that crude terrestrial approximation does not partake of It."
Perhaps the scientist will just bite the bullet and deny science has a true substantial form. Instead, he may say, à la Quine, Hempel, Carnap, Ayer, et alii, that science is just "S(a,o,s): a nominal arrangement of axioms, operations, and conventional symbols." In that case, however, either no genuinely scientific discoveries can break the completeness of that system––rather like a borrowed foreign word cannot violate the larger syntactical and grammatical parameters of the borrowing language––or real discoveries wholly beyond the parameters of that set will destroy 'science' qua S(a,o,s). If the validity of scientific cognition just is its conformity with a conventional axiom-operation-symbol set, then any new line of inquiry will either never count as science or will so radically diverge from S(a,o,s) as to falsify S(a,o,s) itself. Science, then, will be either iteratively derivative or inherently self-destructive. If science is just a Kantian enterprise tracking the ineluctable contours of our own cognition, then no discoveries can actually transcend––'surprise'––our immediate cognitive system qua S(a,o,s).
Objection: Kantian science does allow for cognitive novelty, since the ongoing immersion of the mind in the infinite pursuit of noumenal forms triggers new insights within the cognitive system itself. We are surprised by our own insights. This objection only shifts the problem back one step, since it presupposes a substantial form of the world which generatively constricts and amplifies our cognitive 'inner' world. Hence, if science does have a universal, substantial form, then it demands the existence of such things in general.