Your figures are radically different than the two of the three values given in the scientific sources that I found for solubility in ethanol. (Not the same solvent, but generally it acts pretty similarly.)
A possible cause, though not necessarily so, is that even with the processing you already did, the material was in fact a mixture of some sort. If that were the case, it might be the additional substance or substances that required all the extra solvent.
Something indicating this might be the case is that if most (more than 50% of a substance) is dissolved in 50 mL, then less than 80 mL will be required to dissolve all of it. If 115 mL was required for all of it, then no more than about 1/3 of it could have been dissolved in 40 mL. Now it could be that you mistook “1/3 dissolved” for “mostly dissolved” but it’s also might be that you did not, you were quite right in your observation.
At this point the question seems unresolved for sure, but it seems to me:
A) The pharmaceutical value of ~750 mg per mL of ethanol is probably wrong. It did seem a surprisingly high figure.
B) Your findings would agree with the Sigma-Aldrich value of 10 mg/mL, which seems a surprisingly low figure, but only if your impression that most was dissolved with 40 mL was wrong.
It would be possible to find out if you in fact have a mixture, if you could repeat the process but this time dissolving in only the 40 mL, filtering, and then evaporating the solution and weighing the crystals.
If amount dissolved was in higher proportion than 1 gram in 115 mL – in other words if you recover more than about 350 mg – then you have a mixture, unfortunately.
Actually we really would want to use enough methanol such that whatever ratio of methanol/water is used would fully dissolve the testosterone propionate. So, and I should have realized this before, really the big question is how much methanol/water, in whatever ratio is used for the hydrolysis via water being added with the saturated NaOH, is needed to dissolve the testosterone propionate.
By the way, and now we are really getting into an area where commentary without having done the procedure becomes really lacking, there’s a potential major problem in doing the hydrolysis in methanol/water, particularly where the amount of water is pretty small.
If the nucleophilic attack – fancy phrase for “thing attacking the ester” – is with methoxide, as is likely in a concentrated methanol solution with little water, your immediate product is not sodium estradiol, but estradiol methyl ester. Which is not water soluble, and the whole idea is to convert the EB to water soluble form.
True, maybe sooner or later an attack will be by hydroxide and thus we actually get sodium estradiol. However if we have to wait through a large number of attacks of the wrong kind before getting one of the desired kind, that increases chance of an attack that we don’t want occuring on the TP in that time.
Two possible answers to that:
First, if the planners of this method thought they were getting sodium estradiol, for the most part they may be wrong but nonetheless methyl estradiol may be more soluble than estradiol benzoate so at least something is accomplished, though not as much as hoped, yet still maybe it is OK.
Or second, we could clean up the problem by using a non-nucleophilic solvent, for example acetonitrile/water.
But one really would have to experiment with that to say how to do it.
Getting back to the main problem, if you could repeat the experiment with the 40 mL and see how much of 1 gram does dissolve into that, it would let you know if your 1 gram is at least largely one material, or is very much a mixture.