APPENDIX 149 



Preparation of Curves — Measuring of specimens is now done almost exclu- 

 sively with the standard measuring engine, a modification of the cathetometer 

 method described in Climatic Cycles and Tree Growth (Douglass, 1919, pp. 

 58-59). A subordinate scale on the instrument checks decade sums and 

 eliminates serious errors in measurement. 



Many growth curves show trends due to aging of the tree, sudden favor- 

 able or unfavorable change in environmental factors and the like, which are 

 meaningless in terms of cycles, and hence are removed by standardizing. 

 This process is accomplished by transforming the actual measured values 

 to percentage departures from a mean or standardizing line (run through the 

 data by estimation) containing the trends it is felt should be removed. It 

 has been found that standardizing lines drawn by Dr. Douglass when com- 

 pared with lines through the same data drawn by the writer differ in only 

 a small degree; however, it is evident that even a relatively large difference 

 will have practically no effect on the lengths of the cycles found and only 

 a small effect on the amplitudes. This is particularly true in light of the 

 desirable practise of keeping the standardizing line as free from bends or 

 "kinks" as the data permit. 



Cutting Line — After a curve is smoothed, it is transferred by carbon tissue 

 to a strip of brown opaque paper called cycleplot paper, some 48 inches long 

 by 4 inches wide, and therefore slightly larger than the window of the cyclo- 

 graph. A base line is then drawn on the curve; this is called a cutting line, 

 and the maxima that it isolates are cut out with a razor blade. The present 

 practise of the Laboratory is to use a cutting line that approaches or is a 

 straight base line, usually parallel to the horizontal axis and at the approxi- 

 mate height of mean minimum. On the margins of a deep minimum, the 

 straight cutting line is slightly modified downward, to avoid displacing the 

 center of gravity of the maximum. Those minor maxima which lie below 

 the cutting line are taken care of by cutting out only the very tops; those 

 minima which lie above the cutting line are extended downward by means of 

 narrow tongues. Tests show that cutting lines of different observers yield 

 no definite difference* as far as cycle lengths are concerned ; the amplitudes of 

 those cycles if determined optically from the cycleplot would, of course, be 

 affected by cutting lines departing from the usual procedure. At present 

 amplitudes are obtained by arithmetic summation of the data and are thus 

 independent of the cutting line. 



We may therefore conclude from a study of the errors in the cycleplot 

 building process that, when the standard methods of the Laboratory are 

 used, the resultant extraneous effects introduced into the derived cycle ele- 

 ments are too small to be of any consequence. 



The most important of the errors involved in the actual manipulation 

 of the instrument is the occasional slight change in the focus correction due 

 to the convenient use of cords connecting the focusing device to the carriage 

 on the track above. The practise in cycle analysis is, however, to test the 

 calibration both before and after every analysis or short series, with the 



