22 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 94 



of the table are left incompletely filled/" Owing to influences aside 

 from the periodicity in progress of being computed, the variable under 

 investigation may be particularly high or particularly low throughout 

 the intervals of the time represented by the first fine or the last of 

 the table. To use a part but not the whole of such a line in a short 

 table must produce distortion of the averages. Yet the total intervals 

 given under caption 1 1 are so short that one can ill spare any part of 

 them. The best course seems to be to fill the fivst and last lines of the 

 tables by extending the table a little past the limits set by equality of 

 sun-spot activity as represented in the caption ii. Yet this may also 

 lead to a distortion of the curve of averages owing to changes of phase 

 produced by changes of solar activity. Perfection under these cir- 

 cumstances is unattainable, and some indulgence to irregularities is 

 to be given on these accounts in criticizing the results. 



14. Berlin, Germany. Departures from Normal Temperatures 



It will be difficult, within the allowable limits of tabular and graph- 

 ical illustration, to demonstrate the findings of this research so thor- 

 oughly as to lead the minds of readers to conclusions such as im- 

 pressed themselves on those of us who followed all the computations 

 from day to day. 



a. segregated with reference to sun-spot NUMBERS 

 A. The ii-month Periodicity 



Recalling that, owing to the smoothing by 5-month traveling means, 

 the /-month and 8-month results must necessarily be unrepresentative, 

 let us take up first of all the ii-month analyses. In figure 11 are 

 given all of the ii-month mean curves for Berlin temperature de- 

 partures obtained by the process outlined above in captions 4, 5, 9, 

 10, and II. In order that the reader may more vividly grasp the 

 nature of this work the periodicity computation for low sun-spots 

 for the interval January 1811 to July 1815 is given in table 4. To 

 avoid printing numerous decimal points, the values as given are the 5- 

 month smoothed departures from normal monthly temperatures ex- 

 pressed in tenths of degrees Centigrade. 



Table 4. — Sample Computation of ii-month Periodicity 



16 21 16 IS 12 7 7 3 



-21 —18 —22 —20 —14 —10 —13 —23 



-14 I 3 — 5 — 8 —13 —16 —14 



-29 —24 —32 —30 —13 — 8 —17 —14 



-8 — 2 — 2 — 3 6 — 3 — 9 —13 



-II. 2 — 4.4 — 7.4 — 8.6 — 3.4 — 5.4 — 9.6 —12.2 



In table 3, what is here called a line is there a column. 



