APPENDIX 145 



wheel. This wheel turns by means of a gear meshing with a toothed runway, 

 which in turn moves at a rate that is a small ratio of the motion of the mirror 

 carriage. The resulting motion of the objective is then such that the change 

 of object distance is exactly counterbalanced and the focal plane remains 

 fixed. 



OPERATION 



Calibration — Periods in the cycleplot may be made to give a horizontal 

 alignment of dots in the pattern by the simple process of changing the distance 

 of the movable mirrors. With a series of artificial curves having regularly 

 spaced maxima plotted on the normal scale of a unit every two millimeters, 

 it is possible to calibrate a scale marked on the carriage runway so that the 

 cycle length corresponding to any horizontal alignment may be read directly. 

 The cy olograph is calibrated from 5.0 to 42.0, the smallest division being 0.1 

 unit. 



Nature of the Pattern — If the cycle present in the data is not exactly periodic, 

 that is, with all the maxima equally spaced, then since there exists a mean 

 length of the cycle there will be a mean alignment in the pattern. Depar- 

 tures from this mean horizontal position of the dots indicate a slight lag or 

 advance in the occurrence of the maxima concerned ; hence, the term differen- 

 tial was at one time applied to the pattern obtained. The closer the cluster- 

 ing of the light-dots about the straight line, the closer the cycle approaches a 

 perfect cycle or periodicity. It is analogous to the clustering of dots about a 

 regression line in a scatter diagram as a measure of correlation, though the 

 cyclogram pattern itself is more closely akin to the O-C diagram (in which 

 positions of the dots along a horizontal line represent their theoretical or 

 computed places, and the vertical displacements from this line the observed depar- 

 tures from exact periodicity) } It is evident that a change of period in any 

 interval will be represented by a different trend in the corresponding interval 

 in the pattern, demanding a different position of the mirrors to bring that 

 interval into horizontality. Logarithmically varying cycles will thus form a 

 curved line pattern. 



Procedure in Analysis — Two complete analyses of each curve are made, one 

 with the movable mirrors at successive settings from 5 to 42 units, and the 

 other with the mirrors returning from 42 to 5. The items usually noted are 

 cycle length, duration (dates of beginning and ending), emphasis or cycle 

 strength, fractionization or splitting of cycle into one or more series of sub- 

 ordinate crests, phase changes, changes in length or amplitude and any 

 special characteristics. 



The matter of cycle emphasis needs particular consideration. The im- 

 portance or strength of a cycle is indicated by underlining the figure repre- 

 senting the cycle length, very good cycles receiving two or even three under- 

 lines (see page 147). The factors determining the assignment of underlines are 



1 In the cyclogram, departures from exact periodicity are also indicated by displace- 

 ments from a horizontal alignment, but maxima are plotted in their observed places in the 

 time scale. 



