CORRELATION OF DISTRIBUTIONAL FEAUTRES. 391 



the particular climatic intensity in question causes the limitation of the 

 plant or vegetation in question. The probabilities to be inferred from 

 such a correspondence are extremely strong, but we may place full 

 confidence in such a deduction only when we are sure that no other 

 factor, intensity, or combination of factors, also follows the given 

 vegetational limit. 



It frequently happens, particularly in the United States, that a 

 rapid geographic change in a given climatic condition is accompanied 

 by changes in other conditions, and that the isoclimatic lines for one 

 feature of the climate are parallel to those for another. There is 

 also a frequent reciprocal relation between factors, one approaching 

 lower values as the other approaches higher. All of these circum- 

 stances make more difficult the task of correlating vegetation and 

 climate. All isoclimatic lines which refer to temperature conditions 

 run, in general, in a west-east direction, and in the United States 

 those referring to moisture conditions run mainly north and south. 

 This makes it easy to distinguish at least the temperature controls 

 from moisture controls, however difficult it may remain to discover 

 which of several temperature conditions, or of several moisture con- 

 ditions, may be the most critical in a given case of distribution. 



Wlierever the method of correlation breaks down or is inconclusive 

 in its evidence, there is an opportunity opened for more intensive 

 study of correlations with reference to critical locaHties. It might be 

 possible to secure the most conclusive and logically precise knowledge 

 of the critical factors for plants by using the experimental methods of a 

 physiological laboratory with equipment for the control of conditions. 

 It is, however, much easier to determine the optimal points for a plant 

 by laboratory methods than it is to determine the limiting points in 

 the scale of conditions. It is relatively easy, too, to subject a short- 

 lived plant to experimentally controlled conditions, although it is 

 difficult to give it conditions — ^particularly of soil and light — similar to 

 these occurring anjnvhere in nature, so that the results will have a 

 definite ecological bearing. Perennial plants may be experimentally 

 investigated with respect to their early life-history, or with respect to 

 individual phases of activity, but there are many problems in con- 

 nection with their physiology which demand field experimentation, 

 or the measurement of the uncontrolled conditions of the natural 

 habitat, and all problems in connection with their ecology demand it. 

 The study of the optimal and limiting conditions for vegetation, as 

 contrasted with individual species, appears to be quite beyond the pale 

 of experimental methods and must be carried on by means of instru- 

 mentation and correlation. 



The rather crude cartographic method of correlation that we have 

 used is adapted only to general studies covering large areas. The 

 same method might well be used on a more refined scale for a study of 

 correlations over any topographically simple area for which there was 



