1. The rate of decay remains constant.
2. There has been no contamination (that is, no daughter or
intermediate elements have been introduced or leeched from the specimen of rock).
3. We can determine how much daughter there was to begin with (if we assume there was no daughter to begin with, yet there was daughter at the formation of the rock, the rock would have a superficial appearance of age).
Are these foundational assumptions reasonable? Recent findings seem to indicate that though we ourselves have not been able to vary the decay rates by much in the laboratory, the decay rates may have been accelerated in the unobservable past [1].
You've received some good replies already. I'd like to add something.
Assumption one is an assumption as far as geochronologists are concerned, but it is definitely
not an assumption as far as physicists are concerned; they've tested it six ways from Sunday. Sylas recently posted an excellent message on this subject with further details, at
Re: summing up, & one more question. One or two recent results that call this into question don't mean much when stacked up against the literally thousands of results that support it. In theory we could find that there has been significant change in radioactive decay rates, in the sense that any finding in science could possible be overturned; in practice, don't bet on it. It ain't gonna happen. Decay rates are constant.
Assumption two is used only for certain types of radioisotope dating, and is not a basis for the mostly widely used methods. The most popular types of dating used today, isochron dating and concordia-discordia dating, indicate when assumption three is violated; in that situation isochron dating tells us that it can't give us a date, but concordia-discordia dating can often give us a good date even when assumption three is violated. It is faintly possible that once in a great while it happens that a rock gives a false appearance of age when measured by these methods, by random chance assemblage of just the right concentrations of chemicals, but it is not possible that
all or even a large number of age determinations are wrong; that would be like winning the grand prize in the lottery every time you bought a ticket. (The age errors that the YEC's keep coming up with are just about all obtained by improper sample selection and/or inappropriate measurement method selection).
Assumption three is, again, true only for certain types of radioisotope dating. Both isochron methods and concordia-discordia methods, do not rely on this assumption. In isochron methods, the amount of initial daughter is measured as part of the procedure. Again it is faintly possible that once in a great while it happens that a rock gives a false appearance of age when measured by isochron methods, but again it is not possible that
all age determinations are wrong.
There's a good, moderaetly technical, explanation of isochron methodology at
Isochron Dating and some of the prevviously posted references give brief descriptions. I don't know of any good references on cincordia-discordia dating on the Web. "The Age of the Earth", G. Brent Dalrymple, Stanford University Press, 1991 has a pretty good short section on it.
Note that the YEC's love to discuss potassium-argon (K-Ar) dating, because it's simple, it is used, and it does rely on all three of those assumptions. (The helium stuff mentioned in the OP doesn't have much of anything to do with K-Ar dating). We know that most K-Ar dates are good because they agree so well with other, more robust methods (including methods not based on radioactive decay). K-Ar dating is still used today becasue it's low-cost and pretty reliable, but the majority of dating sutdies are done with other methods, and most K-Ar dates are not published without confirmation by another method or other evidence.
Some more info and links at
Radioisotope dating links and information and
A Radiometric Dating Resource List.