( 316 ) 



ui 



In the second column is given under ' '„ (calculated the distri 

 tion derived according to the simple exponential law; in the fourth 

 column we find the values of the second term in the series having 



Z), as factor; from the third and fourth columns is evident that the 

 sum of the differences is lessened by this term from '215 to 118 a 

 thousand. As has been noticed before, the influence of the third term 

 with D i is slight. 



The sum of the differences remains 12%, also after introducing 

 this term, which can be called satisfactory considering that the total 

 number of observations is not more than : 



2 X 31 X 23 = 1426 



and that the most unfavourable months have been taken as an example. 

 In fact, from the regular course of the differences it is evident thai 

 there might be a possibility of making the differences smaller still 

 by addition of a fourth or a fifth term with D s and D„. 



For D b we find the value — 0.00036, from which ensues that of 

 liv far the greater part the differences are due to incompleteness of 

 the material of observation, so that extension of the series would 

 avail but little. 



4. As fitting material for a second application of the method to 

 meteorological quantities all the observations of temperature have 

 been chosen, taken six times a day in the month of July on the 

 same lightship during the years 1882 — 1906. The number of obser- 

 vations is now six times greater than for the daily-means and amounts 

 to 4516. 



On account of this greater number the frequency-curve will have 

 a more regular shape and the obliquity which was easily discernible 

 for the daily-means also for the summer months, will now come 

 more clearly to the front. 



The -observations are arranged according to the different quarters 

 of the wind, so that we obtain (Table IV) frequencies of the so-called 

 thermic windrose. On board the lightships the direction of the wind 

 is determined in accordance with the indications of the compass; 

 for the period 1882 — 1906 we can assume that these observed, direc- 

 tions of the wind can lie reduced to the proper direction by applying 

 as collection the mean deviation, — 15°. 



