520 A UNIVEKSAL METEOROGRAPII, 



Such an observatory, if it occupied but little space and were suffi- 

 ciently light, might easily be attached to a captive balloon, the cable of 

 which would serve to maintain the required communication between 

 the instruments and the ground. 



The problem proposed by the society is a little vague in its terms, so 

 that it is necessary to repeat it in a more definite form, specifying differ- 

 ent conditions which would materially affect the solution. 



1. A primary and very important point is whether by "considerable 

 elevation" is meant a fixed accessible point, such as the summit of a 

 mountain or a rock in the ocean, furnishing a solid base fur the instru- 

 ments, or an elevated point in the atmosphere to which the instruments 

 are borne by means of a kite or a captive balloon. In the latter case, 

 in fact, a mercurial barometer could not be used, even if hung on gim- 

 bals, since excessively great oscillations would be produced by the 

 slightest breeze. Besides, barometrical observations will be of little 

 value in determining variations of atmospheric pressure unless it be 

 positively certain that the instrument has constantly remained at the 

 same height above the level of the sea, or it be definitely known how 

 many feet the balloon carrying it has ascended or descended. 



It is true the height of a captive balloon may be deduced from the 

 length of cable unrolled and the angle made by it with the vertical; 

 but on account of possible oscillations, ami of the curve which the cable 

 assumes, this determination will always be very difficult and uncertain, 

 and an error of ten or twenty feet in the height of the balloon will cor- 

 respond to a difference of one one-hundredth of an inch in the height of 

 the barometric column. 



2. It is not a matter of indifference, either, in the solution of the prob- 

 lem, whether the observations are to be recorded where the instruments 

 are placed, and therefore be known only at the end of a long series, 

 or whether it be required that the instruments of the detached observ- 

 atory should transmit their indications regularly to the main observa- 

 tory for comparison with the observations made there. The latter sys- 

 tem, which most meteorologists would doubtless prefer, will naturally 

 require telegraphic communication, which the former could dispense 

 with. 



3. Finally, an essential point is to know for what time the automatic 

 registry should be made without the necessity of visiting the detached 

 observatory to wind the clock or make any other change. The difficulty 

 would evidently not be the same if the apparatus be required to run 

 twenty-four hours, a month, or a year. 



In discussing the question proposed by the Netherland Society we 

 will endeavor to keep in view the conditions we have just named, and 

 we will consequently distinguish two different kinds of detached ob- 

 servatories, namely : 



1. Those intended simply to collect observations for ten days at most, 

 the records of which need not be known before the expiration of that 

 period. 



