Please, sir, I want some more

In many ways oxalic acid is the ideal miticide. It is an inexpensive, organic product that naturally occurs within the hive (and in honey) at low levels. It is very well tolerated by adult worker bees and queens, and is extremely effective at killing Varroa mites.
Oxalic acid has multiple effects against Varroa, damaging their feet and causing a range of significant physiological changes {{1}}. The fact that there are multiple targets means that, in contrast to the single-target chemical miticides, resistance of Varroa to oxalic acid is both unknown and unlikely.
However, oxalic acid is not perfect {{2}}; in solution it damages open brood (larvae and, probably, eggs), and it does not penetrate the cappings covering developing pupae, so only kills phoretic Varroa.
Oxalic acid (henceforth 'OA' as I'll be using the term a lot throughout this post) has been used for Varroa control for almost 40 years, and is regularly applied by beekeepers in midwinter — or, more correctly, a broodless period between late-autumn and early spring {{3}}.
The shortcomings of OA, in particular its inability to kill mites in capped cells, means beekeepers have developed colony management and application methods designed to overcome these problems, allowing OA to be used at other times of the season.
For example, midseason queen caging can be used to enforce a broodless period, making all the mites in the colony phoretic. A single dose of OA should then kill 90% or more of the mites {{4}}. However, this type of management method may not be suitable at latitudes with shorter beekeeping seasons, or particular honey harvests.
An alternative strategy, if the colony is not broodless, is to make repeat applications of OA. The timing of these is regularly discussed, but the dose often is not, despite the fact that dose and timing (and the duration of treatment) together determine the success of the method.
This post considers why dose and timing are related and significant, and the scientific studies conducted that show what works … and what does not.