144 THE CLIMATIC FACTOR AS ILLUSTRATED IN ARID AMERICA. 



for future identification if other investigators carry tliis work farther. The total number 

 of measurements, excluding trees not used, amounts to 925, which makes an average of over 

 2 for each tree. Some of these were discarded, leaving 785 as the number actually used. 



As a rule three measurements were made only in cases where the first two differed by 

 more than 1 per cent. More than three were made only upon trees of exceptional difficult3^ 

 The age of the trees at the time of cutting varied from 250 to 3,210 years. The average is 

 approximately 1,400. Including the 28 measurements of the 1 1 discarded trees, the number 

 of individual decades measured during our two seasons of work amounts to approximately 

 111,700, while the number of rings actually counted was ten times as great, or 1,117,000. 

 So large a number of measurements ought to give results of fairly high accuracy. Inasmuch 

 as the number of trees which go back 2,000 years is 79, considerable accuracy is possible 

 at that remote time; and even at a date 600 years earher, where 10 trees are available, 

 the results are sufficiently accurate to indicate the main outlines of the climatic curve, 

 although the vagaries of individual decades are not reliable and the general fluctuations 

 are exaggerated. Finallj^ the data used in the curve are derived from an area sufficiently 

 wide to eliminate in large measure the effects of purely local phenomena. Most of the 

 trees grew within a circle 10 miles in diameter centering between Hume and the General 

 Grant National Park, l^ut a considerable number came from the Tulare region, where they 

 were distributed in two tracts 8 miles or more from one another, and 60 miles from the 

 main area. These trees from the Tulare region give curves whose main outlines are the same 

 as those from the other area. Therefore it seems that local circumstances other than varia- 

 tions of climate have not had any noticeable efTect upon the main form of the final curve. 



In order to obtain exact results it is necessary not only to have a large number of 

 measurements, but to determine all possible sources of error and make alowances for 

 them. In addition to the corrections for age and longevity, which have been described 

 in the preceding chapter, three other types of correction are needed. The first of these 

 is a correction to ofTset the errors in counting, which are inevitable if the counting must 

 be done by fallible human beings. Next a correction is required to offset the fact that the 

 trees, as has already been said, do not always form complete rings each year, and so in 

 some cases seem to be younger than they really are. Finally, the fact that the trees flare 

 at the base, together with the fact that it is necessary to select the best rather than the 

 average radii, demands still another type of correction. 



In order to find out how large a part the individual idiosyncrasies of the various 

 observers have played in causing errors of counting, I instructed my assistants to make a 

 recount of a number of measurements. Unfortunately I did not reaUze the necessity of this 

 until after I had left tlie field, and so my own error and that of four assistants, each of 

 whom made a small number of measurements, could not be determined. For the three 

 assistants, however, who with myself did four-fifths of the counting, and measuring, the 

 statistics are sufficient to show the nature and degree of the errors involved. The method 

 was simple. After one observer had finished a radius, he was instructed to go over it again 

 and count the number of rings, and his two fellow-observers were instructed to do Hkewise. 



Table 4 shows the result. In each case the number of radii involved is 46 and the aver- 

 age age is 1,472 years. The first column shows the age of the trees according to the original 

 count. The other columns show the extent to which the recounts of the three observers, 

 X, Y, and Z, differed from the original. It will be seen that in 22 cases, or nearly half, 

 X obtained the same result as in the original count, which in many cases was made by 

 himself. In 16 cases he obtained more than the original, his greatest divergence being 

 61 on a very difficult tree with small, almost invisible rings, and his average divergence 

 being 10.1. In the remaining cases, 8 in number, he got less than the original, his worst 

 result being -47, and his average -12.6. His divergences on the plus side exceeded 

 those on the minus side by only 60, so that his average difference from the original was 



