Every letter we have printed

Your worked example separates two species one dalton apart and treats that as the hard case. The hard case in practice is fourteen milli-daltons, which is what distinguishes a genuine methyl group from a substitution nobody wanted. That gap needs an order of magnitude more resolving power than most service instruments deliver.

Resolving power and mass accuracy are independent and your article treats them as one virtue. A time-of-flight instrument can be well calibrated and modest in resolving power; an ion trap can resolve and drift. The two specifications answer separate questions and a report should carry both.

A small defence of the linear MALDI instrument. It is fast, it tolerates dirty samples, and for a synthesis chemist checking that a chain has grown by the residue intended it is entirely fit for purpose. The problem is not the instrument. It is printing its output on a release document.

Water is the usual source and glassware is the usual carrier. Depyrogenation of the container by dry heat is the step that gets skipped in small operations, because it looks like a detail and it is the difference between a clean result and a puzzling one.

A note on presentation. Where a value is below the limit of detection, the report should give the detection limit, because "not detected" with a high limit and "not detected" with a low one are very different statements.

The test is destructive, which is the point your piece makes lightly and which deserves more weight. Every unit tested is a unit that cannot be sold, and on a small fill that is a real proportion of the batch. It is an argument for process assurance rather than for testing more.

A supplier that started publishing this data would differentiate itself immediately, which suggests either that the cost is higher than it appears or that nobody has thought of it. Both are interesting answers.

Retitration after an interruption is discussed as though the tolerance acquired during the first course persists. Whether it does is an empirical question and I can find no study addressing it in this class.

Supply interruptions are the commonest cause of the situation your article describes, and they are not a clinical decision at all. It is worth connecting the titration question to the supply reporting elsewhere in the issue, because for many readers they are the same story.

In this market the nominal dose is derived from an assumed concentration, so the real ceiling question is a question about content. A figure calculated from a label is not a dose in the sense the trial literature uses the word.

Divergence between practice and label is used as an argument in both directions and supports neither. That people do something differently is a fact about people, and it carries no information about whether it works.

The trial schedules escalated on a fixed calendar regardless of response, which is a design choice for comparability rather than a recommendation. Reading them as optimal is reading a protocol as advice.

The interesting number is not how many people experienced an effect but how many changed what they were doing because of it. The second is collected in most trials and reported in few.

Where a trial permits a dose reduction, the reported tolerability figures describe a population that adjusted, not a population that persisted. It is a reasonable protocol and it means the headline rate is not a rate at the nominal dose.

The most useful presentation I have seen was a cumulative curve rather than a proportion, because it showed both how many and when. It is not harder to produce and it is almost never chosen.

Trial populations are screened, and the screening removes exactly the people most likely to experience the effects being counted. Every incidence figure in this literature is a figure for a healthier-than-average population.

I keep back issues, which I do for no other publication I read, and the reason is that the reference material stays accurate for longer than the news does.

Consider the cost. A genuinely blind programme means buying material at retail, paying full freight, waiting in the ordinary queue and absorbing the whole expense with no cooperation from the seller. That is why so little of it happens, and the reason is arithmetic rather than principle.

Pre-registration would cost nothing here. Say in advance which suppliers and which molecules will be purchased, then publish everything that comes back. It is the single reform that would make the difference between a testing programme and a collection of anecdotes.

Sellers who cannot obtain a badge sometimes make their own. I have three in my collection that imitate a service’s house style closely enough to pass a glance and carry no link at all. The genuine mark has created a market for its own imitation.

Badge and product are separated by a layer of navigation on most sites I visit. The badge sits in the footer, the product is on a different page, and the association a buyer draws between them is entirely their own work.

An honest presentation would give the ratio with its confidence interval and the sample size in the same sentence. Three numbers instead of one, and the impression created would be entirely different.

Age and sex distributions in the substudies differ substantially from the populations the figures are quoted about, and composition change is strongly patterned by both. Anybody reading a single ratio across a general population is discarding the two largest sources of structure in the data.

Small correction to your table: the S-LiTE exercise prescription was two supervised group sessions and two individual sessions weekly, not two sessions in total. The distinction matters because "add some exercise" is not what was tested.

The word discontinuation implies an endpoint that has been reached. A great many of the accounts I read describe an intention to resume that was never acted on, which is a different phenomenon recorded under the same heading.

Loss of confidence in the material is a reason your taxonomy does not include and I have seen repeatedly. A batch that behaved differently from the previous one ends more arrangements than any pharmacological consideration, and it is a supply-quality question rather than a clinical one.

Nobody discusses what maintenance does to the appetite for other interventions. Several accounts I have read describe a period in which nothing else was attempted because the current arrangement was working, and the cost of that pause only becomes visible later.

The phrase lack of efficacy conceals two different things: no effect at all, and an effect that stopped being worth the effort. They have different implications and they are recorded under one heading.

Nobody separates the reason for the first stop from the reason for the last one. People who stop and restart several times have a pattern, and a pattern is more informative than any single event within it.

You describe the maintenance strategy of stepping down one dose and holding for eight to twelve weeks as something clinicians report doing, and then say it is not a recommendation. That distinction will be lost on most readers, and printing the protocol makes you a source for it whether you intend to be or not.

Tapering is described in the community literature with a confidence the published evidence does not support. There is no trial that compares a taper against an abrupt stop in this class, and the absence should be stated wherever the practice is discussed.

A long half-life means every stop is a taper in pharmacokinetic terms, whether or not the dose is stepped down. The concentration declines over weeks regardless, which is a point worth making before any discussion of tapering schedules.

Duration is the open question underneath all of this. If the honest answer is that nobody knows how long is appropriate, that should be the sentence rather than a set of conventions presented as guidance.

A limit expressed in endotoxin units means nothing without a per-kilogram basis and a route, which is where the specification actually lives. Numbers quoted in this market are usually bare figures with no denominator at all.

A result on one vial from a lot is a result about that vial. Endotoxin contamination is frequently sporadic rather than uniform, which is precisely why pharmacopoeial sampling plans specify more than one unit.

Coring is the failure mode that follows repeated withdrawals from one stopper, and it produces exactly the particulate your article warns about. The number of entries a closure is qualified for is a real specification and it is never stated on anything sold in this market.

A point from the inspection side. Terminal sterilisation, where the product tolerates it, is a different and stronger proposition than aseptic processing, and peptides generally do not tolerate it. That is the real reason this whole subject is difficult, and it is a chemistry constraint rather than a corporate one.

Nothing in this area has been studied in the population that most reads about it, which is people acquiring material outside a clinical setting and training without supervision. Every figure is a read-across and should be labelled as one.

Adherence is the variable that does most of the work and it is the one least often reported. A programme that is performed twice a week for a year beats a better programme abandoned in March, and the trial literature systematically studies the second kind.

A reason absent from every list: the person stopped believing the material was what it claimed to be. In a market where that is a live question, it is an obvious category, and no clinical taxonomy would ever include it.