F. Dellmann on Atmospheric Electricity. 415 



AVith the electrometer here employed the last precaution is ne- 

 cessary, even in order to compare later with former numbers in 

 the case of the glass thread being destroyed. Coulomb's for- 

 mula, it is true, might serve this purpose; yet Kohlrausch's 

 proposition to use the tension of the pile for this purpose is pre- 

 ferable, since the method is more practical. In the following 

 numbers, therefore, the unit is the tension of an element of a 

 zinc-copper battery. With respect to the reduction, the follow- 

 ing is the best method. 



A battery is constructed as before described*, and the tension 

 of its several parts is measured through several days or even 

 weeks. The quantity given by the Table for each angle of de- 

 flection divided by the number of elements, gives of course the 

 quantity for one element. The mean of several hundreds of 

 these numbers is used as a divisor for all the numbers of the 

 Table, and the corresponding quotients taken to form the Table 

 for this glass thread. The observations may be very conveni- 

 ently calculated by means of this new Table, especially if it be 

 provided with the differences between the successive magnitudes, 

 in order to find quickly, for fractions of an angle, the correspond- 

 ing additions to be made to the quantities corresponding to the 

 integers contained in these I'espective angles. 



At our station the atmospheric electricity was measured regu- 

 larly three times a day, at the hours fixed by the Royal Meteor- 

 ological Institution, that is to say, at 6 a.m., 2 p.m., and 10 

 P.M., and, in addition, as often as the study of the phsenomena 

 rendered it advisable, and the engagements of the observer per- 

 mitted. Two measui'ements were generally made each time ; 

 but if the angles of deflection of these differed by more than 3°, 

 the measurements were repeated until two deflections at least 

 lay within" the above limits. In the case of negative electricity 

 alone, which generally causes great deflections, and is not taken 

 into the whole calculation, one measurement often sufficed, par- 

 ticularly when the quantity was considerable, for then no great 

 difference in the angles was to be expected. Where more than 

 two measurements were made, the mean of the two angles f which 

 differed least from each other formed the magnitude to be cal- 

 culated. On the whole, considerably more than 4000 measure- 

 ments of this kind were made last year. 



The monthly means may be here first given J : — 



* PoggendorfF's Annalen, vol. Ixxxvi. p. 535. 



t Strictly, the mean of the quantities should have been taken ; but, as 

 Kohh-ausch has proved, the error thus committed lies within the limits of 

 errors of observation. 



X These monthly means agree better with those communicated by 

 Schiibler in his 'Meteorology' (p- S5), as results of two years' observa- 



