I found this article in a PDF of a booklet published by the Australian Skeptics Society. Just to be clear, it was not written by me, but by Professor Andrew Gleadow. I think it is the simplest, shortest, and best summary of Radiometric Dating so far. The only thing that he did not go into was the constancy of decay, which of course is discussed elsewhere.
How Long Has the Candle Been Burning?
By Professor Andrew Gleadow
It is quite true that in order to estimate the total time that the candle has been burning you need to know not only the rate at which the candle burns, but also that the rate has been the same ever since burning began, and exactly how much of the burnt material was there to begin with. This is indeed analogous to the basic information that is required for geological dating techniques, as is pointed out clearly in any textbook on the subject of geochronology.
What is misleading, however, about the argument presented opposite is the implicit assumption that the behaviour of candles during burning is somehow so mysterious that it cannot be understood, or that it may not follow known physical laws. It is quite obvious that the laws controlling the burning behaviour of candles can quite easily be determined by experiment and observation. If candles of this type are always observed to have started at a certain length and to bum at a constant rate, then they could indeed be used as a kind of clock. Candles have, in fact, been used in the past for this very purpose. It is simply not true, and quite illogical, to suggest that the burning of candles cannot be understood well enough to estimate how long a candle has been burning. In using natural radioactivity to determine the ages of rocks, many careful experiments are conducted to determine the physical behaviour of the particular measurement systems being used. The behaviour of different dating systems can be investigated directly by experiment and observation to determine, for example, the amount of daughter isotope (the "burnt material") that might be included at the time of formation. For radioactive decay rates to vary through time would violate the known laws of physics, meaning that all science would have to be wrong, not just a few "inconvenient" rock-dating measurements.
No guess-work is required in determining how much "burnt material" was present initially in the radioactive dating systems used in geology, which are based on the measurements of a radioactive element and its decay products, or the accumulated effects of the decay. A number of techniques are available to determine whether any of the product material was actually present when the system began. These include the isochron method, which requires no assumptions about how much daughter product was present initially, or using mineral systems which are known experimentally to incorporate no daughter product when they are formed.
Another approach is to look at the isotopic composition of the daughter product in minerals in the same rock which do not contain any of the radioactive decay element.
Radiation-damage techniques, such as fission-track dating, study the damage produced in natural materials by radioactive decay. Such materials clearly cannot incorporate radiation-damage from before they were formed. In geologically undisturbed systems, all of these approaches can be shown to give the same results indicating that many rocks, and therefore the Earth itself, are of enormous age (billions of years). The answer to the argument presented is simply that it is possible to understand natural systems and physical laws. Such understanding leads to predictions that can be tested to the point where the underlying physical laws cannot reasonably be doubted. On this basis, it is entirely possible to use a candle to estimate elapsed time, and the same is true for geological dating systems.