NO. lb TREE GROWTH AND RAINFALL CLOCK 5 



METHODS 



Because all cores consisted of sound wood, none was discarded. 

 Furthermore, because site factors such as light, drainage, slope, 

 ground-water relations, and competition were evaluated on the spot 

 as closely as possible, no reason existed immediately after the col- 

 lection had been made for the rejection of any specimen. The collec- 

 tion was considered a normal representation of the site factors at 

 the three chosen localities even though the sequences dififered to a 

 great extent in variability and average growth-layer thicknesses. At 

 the time the cores were taken there seemed to be no reason why 

 different species should show differences except those due to slight 

 variations of site factors peculiar to each tree. Such a factor as 

 soil aeration had to be judged by soil texture and composition and 

 visible soil-water relations. There was no opport-imity for analyses 

 or measurements. Indeed, this problem of selection in the field, with- 

 out measurements, was of great importance: could local site factors 

 be judged with sufficient accuracy to demand the inclusion of each 

 core as a representative specimen in the general collection? If so, 

 choice in the field, based on ecologic principles, would be a dependable 

 method of selection whose integrity could be questioned only on field 

 evidence or its derivatives. 



Treatment of the wood. — The cores and the growth layers they 

 contained were subjected to the following procedure to prepare them 

 for correlation among themselves and with rainfall. 



1. The cores wxre glued in a groove sunk into the curved side of 

 half-inch half-round and "shaved" by razor sufficiently to expose the 

 growth layers clearly.^ 



2. Beginning with the increment for 1946, which was complete be- 

 cause of the time of sampling, October 5, 1946, the growth layers 

 were counted inward and dated on the assumption that each sharply 

 bounded layer represented a year. 



3. Skeleton ^ plots were set up on coordinate paper, each ordinate 

 representing a year. If a sharply bounded growth layer was decidedly 

 thinner than its immediate neighbors an ink line was drawn on the 

 ordinate appropriate to its date, the height of the line being inversely 

 proportional to the thickness of the growth layer. The resultant 

 skeleton plots and the master plot derived from them are shown on 



5 Principles and methods of tree-ring analysis. Carnegie Institution of Wash- 

 ington Publ. No. 486, p. 6, 1937. 

 «Ibid., pp. 14-16. 



