Question from a non chemist: Ok so you do some chemistry and produce a vial of supposed medicine, how do you tell that you didn't fuck up and produce something toxic? I guess a modern lab would just stick the result in a gas chromatograph or something but you're not going to have one at home.
I vaguely recall (from, er, Sherlock Holmes) that old school analytic chemistry could figure out what an unknown chemical with cheap reactions and tests, but does anyone even know how to do that any more? Is it automatable?
I advise anyone interested in pharmaceuticals to read The Old Vegetable Neurotics: Hemlock, Opium, Belladonna and Henbane, Their Physiological Action and Therapeutical Use Alone and in Combination, from the 1860s, I think. If you read between the lines, it is the story of a doctor who is immensely frustrated with the impurities of some common medicines, not knowing what you were getting, was this the actual chemical doing the job, and so forth. He proceeds to examine each part of the plant (leaves, stems, roots, flowers, and so on) to see which gives the best concentration. He then works at extracting exactly what he wants, very carefully. Then he tests tiny amounts of various rodents, then his dog, then himself, working to determine an appropriate dosage.
It's quite impressive and hardcore. You can feel this man's annoyance at the state of things and his passion to declare, "I'd rather just find this out now instead of waiting for everyone to keep fiddling about." With a small room and some glassware you could also do this. Now, it wouldn't be simple, and to be honest the various units used are a hot mess, but the basic principles are there.
Now, automatable is a matter of degree. Real chemists have whole flowcharts of "do this, do that next, now this other thing" to detect the presence of a particular sought-for thing, but that does not prove a negative, you do not know what else might be present. That's the real rub and why the focus of so much chemistry is on extraction and purification.
The old school way was basically try a reaction, see if it matches a group of chemicals. Try another, see what you get. You can only close in on a reasonably accurate guess if what you're working on is fairly simple. Otherwise you'll be stuck in a local maxima of has this group, doesn't have that group, without the ability to get any better resolution.
I vaguely recall (from, er, Sherlock Holmes) that old school analytic chemistry could figure out what an unknown chemical with cheap reactions and tests, but does anyone even know how to do that any more? Is it automatable?