It is generally believed that genetic drift occurs as a result of sampling error. As we said earlier it occurs in small populations such as peripheral isolates. We may demonstrate genetic drift by a small experiment. Take beads of same size but different colours, say blue, red, green and yellow. Take a thousand of each colour and mix them well in a bag or mug. The 4000 beads now constitute a population. Now put your hand in without looking, pick up just four beads with your finger tips. Let us say that you get two blues, one red and one green. This will mean that the blues have increased from 25% to 50% but the yellows are reduced to zero. You could see that the random drifting or large scale changes in the frequencies of coloured beads is a result of sampling error. Just as the frequencies of coloured beads drift randomly in the experiment, in small populations or peripheral isolates gene frequencies may drift due to sampling error. Hence the phenomenon is known as genetic drift.

Can such drift be observed in natural populations? The answer appears to be yes. Let us take a small population of mice living in the rice barn of a farmer as four or five extended families. The farmer tries a variety of methods like setting up of traps, use of a shot gun, surprise visits, cats etc, to eradicate them. Such acts of the farmer ,' exert a severe selection pressure on the mice. Under such circumstances the trails that would be selected are the swiftness, short tail, hearirig acuity, cautiousness etc. Naturally the frequencies of the alleles that control thcsc traits would tend to be high in the population as only those mice which possess such traits can survive in an hostile environment. After a couple of months, let us say there is an environmental change and a severe winter sets in. The farmer confines himself to a fire place and as a result the selection pressure on those traits we mentioned earlier is now liftcd. And in order to survive in a changed environment, the mice need to possess totally a different set of traits and essentially mice with such traits would be selected for. Thus swiftness and visual acuity are no more the trails that would be selected for, but the mice with a thick fur on them and similar such traits which would protect them from the severity of the winter would be selected. The net result is that the frequencies of the alleles that controlled the traits in the earlier environment undergo a drift (i.e.) their frequencies become significantly low in the new environment. It should be emphasised that such a drift is characteristic of only small populations.