A colleague and I did a modest search of available information on perborate latensification then mulled it over for a while. Here is our lazy, hurried plagiarism of other people’s hard work, with a few insignificant additions, fantasies and elaborations of mine:

Perborate exists in solution as a peroxide-perborate equilibrium. The latensification effect appears to be caused by the perborate reducing silver halide onto the existing silver atoms in the latent sub-image specks, thus making them developable. Therefore it is advantageous to adjust the pH to favour the perborate: pH 10 is the minimum pH necessary, and pH 11 gives some improvement. Lower pH favours peroxide which is less selective: therefore it produces lower speed gain and higher fog.

Film selection
The faster the film, the less the speed gain.

Spectral response
The action does not involve the conversion of photons to electrons, so is independent of the method of spectral sensitisation – ie it is not spectrally selective.

Latensification will not work if the intensity of the light is low enough to require long exposures – it seems that one second or thereabouts is the upper limit.

Because the latensification action is on image specks with few silver atoms it is important to do it as soon as possible after the formation of the latent image. Storage temperature has an effect on the stability of clusters with few silver atoms. The potential speed gain may be reduced by 50% after a week at 20°C.

2 g/l sodium perborate was effective if the emulsion was dried after treatment and before developing.
10 g/l sodium perborate was as effective, but didn’t need to be dried.

30 seconds at 20°C was adequate. Most of the action occurred during the first 5 seconds. If the film is not subsequently dried, it should be left for a couple of minutes after the perborate bath before development. If the film is dried, additional speed gain is possible by repeating the perborate/drying cycle.

You should carry out tests to determine optimum development time. The longer the film spends in the developer, the less effect the perborate has on the final image. (my thought: does this favour a developing agent like Metol that has rapid shadow action, and a low density range in the negative? )

The perborate does increase the graininess of the images, but it is estimated to be less than the increase in graininess by using faster film.
All the ‘speed increase’ happens in the toe. There is little change in the midtones and nothing in the shoulder.

I'm continuing with my own comparisons, but a very minor leg injury has slowed me down a little, pathetic wimp that I am.