The proposition that a drug placed in the nasal cavity can reach the central nervous system without first entering the systemic circulation has organised two decades of preclinical pharmaceutics, yet the claim remains contested at its foundation. This review argues that the difficulty is not biological implausibility. Anatomical substrates for olfactory, trigeminal and perivascular transfer are real and have been characterised in detail. The difficulty is evidentiary. Direct transport is a claim about route, whereas nearly every experiment that supports it measures only outcome, namely the amount of drug recovered from brain tissue. Bridging that gap requires a specific inferential architecture, and the metrics the field relies upon, drug targeting efficiency and direct transport percentage, are ratio estimators whose validity depends on assumptions about dose equivalence, systemic disposition, sampling adequacy and tissue integrity that most published designs neither satisfy nor examine. The steps that most strongly determine the result, namely instillation volume and depth, animal posture, anaesthesia, the intravenous comparator and the dissection protocol, are also the steps least completely reported. Pooled surveys of this literature have repeatedly found variability in reported targeting that does not track the physicochemical properties of the delivered agent, which is the signature of methodological rather than pharmaceutical variance. Generic reporting frameworks, ARRIVE 2.0, SYRCLE and MIRIBEL, each address part of the problem and none addresses the route-specific part. This review therefore sets out a reporting standard built for intranasal central nervous system studies, organised across administration, comparator, sampling, analysis and translation, together with a four-tier scheme that states plainly which conclusions a given design can carry. Adopting an explicit evidential grammar would allow the field to accumulate evidence rather than merely accumulate publications.