He cited as an example the experiments on the Viking Mars landers, in which a sample of Mars soil was exposed to nutrient broths to se whether they consumed them and produced detectable gases.

This test was generally based on the common belief that life can be defined as regions of "negative entropy", in which matter organizes itself into more ordered forms and expels its former disorder into the outside environment, in the form of such things as temperature changes and chemical changes.

But while negative entropy is a necessary part of the definition of life, it's not adequate -- a lot of nonliving processes (such as crystallization and other chemical reactions) also decrease local entropy, and so the Viking experiments ended up being confused by a wholly unsuspected (and still rather puzzling) set of nonliving chemical reactions in the Martian soil that also produced the gases they had been looking for.

A better definition of life is "self-reproduction" -- and Chao pointed out that the Viking test could easily be modified to look for this. A sample of the soil-nutrient broth mixture from the first test container could be transferred to another container of clean nutrient, to see whether the agent that had consumed the nutrient in the first container was simply diluted in the second container and produced less gas, or whether the agent reproduced itself and ended up eating as much of the nutrient in the second container.

But even reproduction is not entirely limited to life; there are a fair number of nonliving "autocatalytic" chemicals that actually react with other chemicals to produce a steadily increasing supply of themselves and could thus fool such a test.

So Chao proposes another definition of life: something that evolves into new and more efficient forms in accord with natural selection and Darwinian evolution.

In Earth labs, any species of microbe that is put into a nutrient will usually not find that nutrient entirely to its liking -- but microbes mutate very rapidly, and after about a month a new strain will usually have developed that metabolizes and consumes that nutrient more efficiently and crowds out the old microbe.

Thus, if we modify our Mars experiment again and keep transferring a small sample of soil and nutrient mixture from one flask to a new flask of fresh nutrient, after about a month the agent consuming the nutrient will actually start doing so more efficiently -- if it is alive.

Chao calls this the only clear definition of life he can think of -- the only kind of test that completely avoids the danger of "false positive readings".

(He provided one interesting footnote: in the struggle for survival, a species crowds out a competitor not by becoming more fit in absolute terms, but just more fit as compared to its old competitor -- and his researches into the evolution of viruses have shown that this means that newly evolved microbes often crowd out their old predecessors just by stealing and parasitizing useful chemicals produced by the old ones.

But once that old competitor is exterminated, the new strain cannot survive without it and dies out too! He thinks that cell membranes first evolved simply because they partially segregate self-replicating biological molecules from other such molecules and thus prevent this kind of self-destructive competition from getting out of hand.)

It should be clear that the Astrobiology Conference provided a great supply of suggestions both as to how life may exist elsewhere in our own Solar System, and how we may go about looking for it.

But any life we find elsewhere in this Solar System will be microscopic and primitive.

What are the odds that more complex life forms -- multicellular, and perhaps even intelligent -- exist in the solar systems of other stars? In my final report on the Conference, I'll look into this crucial question.