"Richard;
I agree with your observations. You have explained how too many variables can obfuscate the problem.
I have a few more specific examples.
Lets assume a 5% impurity in commercial Borax.
If it is Sodium Sulfate, there will be virtually no effect on pH or on sensitometry if you use standard levels of buffer and dilution, but if you use the developer for long times or dilute it substantially, there is the potential that contrast and speed will go down. You see, since Sodium Sulfate is neutral to this situation, the developer is 5% low in buffer capacity and this will only show up in keeping, heavy use or high dilution. D76 diluted would be a case in point.
If it is Magnesium Sulfate, it will form a solid precipitate which is about 5% by weight. If you filter out the solid you are pretty much back to the above scenario, but if you add enough sequestrant (TSP for example) you will form Magnesium Phosphate which will raise the pH by as much as 0.1 - 0.5 pH units and change the buffer capacity downwards. Phosphate alone is a poorer buffer than Borate at pH values where Borate is used. This can cause an initial increase in contrast which would then drop with dilution, use or keeping.
If you varied between a 5% contaminant of one or the other or just inert silicates, you would have a situtation where you were varying in pH to a small extent and in contrast and speed which might be on the order of 1/3 stop or about 0.05 - 0.1 contrast units. This contrast variation would be on the order of 1/2 grade in the paper you might use.
BTW, this also applies to 'pool grade' Sodium Carbonate which was discussed in the thread Kirk referred to above. The pH and buffer capacity would vary by as much as 5% depending on the nature of the impurity.
I've already explained what would happen if this 5% impurity was a halide, but lets assume that the impurity consists of mixtures of all of the possible (so called benign) items discussed so far. Then the diagnostic process and the results become rather hopeless as Richard described above. It also decreases your process uniformity if you strive for quality results.
The presence of minute silicate particles might not even be noticed if small enough. If large enough, you might mistake them for grain or you might not see them until you reached higher magnification.
This has been a discussion on what you can observe and test for. I have left out the potential effects on sharpness which would be about 5% as well, but much harder to test for the average person.
PE