Solar Magnetic Period. 439 



The mean temperature of each division for each day was 

 taken out ; then the mean temperature for the period from 

 these means, and also the mean temperature from the middle 

 of one period to the middle of the next; from these last means 

 were interpolated values for each day ; finally the differences 

 between these two sets of numbers gives the daily departure of 

 temperature. In this way the annual changes of temperature, 

 due to the motion of the sun in latitude were to a considerable 

 extent eliminated. The final sum of these residuals for each 

 day is entered in the accompanying table. It will be observed 

 that they are divided into two nearly equal parts which was 

 accomplished as follows : The residuals of each period were 

 plotted on half centimeter paper, and the line of variation of 

 temperature drawn through them. These curves were care- 

 fully compared with the mean magnetic curve, which repre- 

 sents the intensity of the solar field, by superposition, when it 

 was easily seen that certain periods agreed with the magnetic 

 curve in its direct position, and others with the same curve in- 

 verted on its long axis. The magnetic curve is transferred to 

 transparent celluloid, so that such comparisons are very easily 

 performed. 



The consideration of these curves disclosed some conditions, 

 which are inherent in this type of work. They are the final 

 effect of a number of causes. (1) The temperature gradients 

 produced by the equatorial field, together with the wind cur- 

 rents by which heat is transported from station to station ; (2) 

 The annual change of the sun in latitude ; (3) The rotation of 

 the sun on its axis carrying the polar magnetic field. The 

 result of these variations is to cause the formations of the indi- 

 vidual curves corresponding to the solar rotation to be quite 

 loosely constructed. The phases may fall a little to one side 

 of the mean date, the amplitude of the crests is not uniformly 

 built up, the mean elevation of the curve changes in an irregu- 

 lar manner, and the inversion of the curve may fall some- 

 where within the period. The magnetic curve itself is one of 

 rapid variation, and hence it is difficult for it to persist steadily 

 under all these sources of disturbance, so that the tendency of 

 the residuals to cut themselves down algebraically must be 

 admitted. The mean curve can be found only by using an 

 extensive series of data, and ascribing to the small surviving 

 residuals a larger degree of importance than at first sight they 

 might seem to justify. These have been regarded as true rela- 

 tive numbers, and the final residuals have been restored to 

 their proper magnitude, which can easily be concluded from 

 the tabulations, by use of an appropriate factor. The residuals 

 are thus to be considered as relative numbers, rather than abso- 

 lute values in themselves, the magnitude of the single curves, 



