96 DISTRIBUTION AND MOVEMENTS OF DESERT PLANTS. 
of winter and summer rainfall at 53 different places, from such records as 
are accessible, shows that, in general, at points east and south of Tucson, 
the aggregate summer rains are considerably in excess of the winter 
rains, while to the west of that place the reverse is true (table 7). This 
difference is due to the fact that to the eastward and southeast the Gulf 
storms become more and more a controlling element, while on approach- 
ing the Pacific the conditions become dominant which, in southern 
California, limit the rainfall to the winter months. This difference is 
such as to very noticeably affect vegetation. Particularly is this true of 
range grasses which by reason of preponderance of summer rains to the 
east and south find there more favorable conditions for development than 
to the westward, a matter of much economical importance to stockmen. 
Table 7, compiled by Professor Thornber, shows how few are the excep- 
tions to the general rule as stated, and that it holds true notwithstanding 
rather wide differences of altitude, which necessarily affect the amount 
of rainfall. 
TEMPERATURE RECORDS. 
The account that has been given (pp. 44-47) of soil and air tempera- 
tures at the gulch near the Desert Laboratory points to differences of 
temperature as the primary factor to which the remarkable differences 
of the plant covering on the two sides of the gulch are due. In this case 
it has been possible to establish a causal relation between observed facts 
of distribution and a single physical factor, which though operating in 
part indirectly is none the less the fundamental cause of the phenomena 
in question. The rather voluminous readings from which the results 
are deduced are given in tables 8 to Io, in sufficient number to indicate 
the nature of the data. Maximum and minimum readings are for either 
one or two days, except in a single instance in which a period of three 
days was covered. 
In field studies of this nature nothing is more important than to bear 
in mind the futility of attempting to account for facts of distribution 
without due consideration of all the factors involved. To emphasize 
the influence of a given factor is one thing; to ignore those acting in 
conjunction with it is quite another. As pointed out by Dr. Cannon (p. 60 
et seq.), the superficially placed root-system of Cereus giganteus subjects 
it to conditions of soil temperature widely different from those affecting 
Parkinsoma nucrophylla, with which, however, it is almost constantly 
associated, and innumerable similar cases emphasize the fundamental 
fact that plant associations, while often conspicuously related, as a whole, 
to a single factor, are composed of species often of widely different habits, 
responding in different ways and degrees not to one, but to all the factors 
of their environment. Under such circumstances any practical method 
of simplifying the problem of measuring and correlating physical factors 
with choice of habitat is particularly welcome. Some progress has been 
made in developing such a method, as will be seen on pages 99-100. 
