So much, then, for relationships between birds and their environments at a descriptive level. It would be useful at this point to examine how environmental variables relate to timing of breeding. Certain independent lines of investigation indicate that birds have a well-developed internal timing device; most convincing is the work of Schmidt-Koenig (1960) and the others who have shown that the endogenous clock of birds can be shifted in its periodicity forward or backward in time. This and much other evidence (see Brown, 1960) indicate that many fundamental periodic regulators are extrinsic to the animal; it is thus permissible for present purposes to consider any expression of variation in timing as dependent on environmental oscillators. It is not hereby meant to ignore the fact that differential responses to dominant environmental variables occur within a species, indicating endogenous control over timing of breeding. The work by Miller (1960:518) with three populations of the White-crowned Sparrow, revealing innately different responses to vernal photoperiodic increase, is especially important in this regard. For the moment, however, we may consider exogenous controls only. Any exogenous control, or environmental variable, can be looked on simply as a timing oscillator. Such variables show regular or irregular periodic activity, and the independent actions as a whole result in the more-or-less variable annual schedule of breeding for any species at any one place. It would seem that some oscillators are linked to one another, but there is a real question concerning the over-all degree to which linkage is present. It is significant that frequency distributions of breeding activity of various species and groups of birds take on the shape of a skewed normal curve. The more information is added to such distributions, the more nearly they approach being wholly normal, with irregularities tending to disappear. This kind of response itself is evidence that most of the variables influencing the distribution are not mutually linked. This conclusion is warranted if we examine what would happen to frequency distributions if the variables or oscillators regulating timing were linked. The frequency distribution of breeding activity in birds is described by a nonlinear curve (a normal distribution is nonlinear). Let us assume that each of the environmental variables is a nonlinear oscillator, as is probable. A set of nonlinear oscillators mutually entrained or coupled and operating with reference to a given phenomenon would result in that phenomenon being described by a frequency distribution much more stable than if it were regulated by any one oscillator alone. However, the frequency distribution of a set of coupled nonlinear oscillators is non-normal (Wiener, 1958).