FOR DETACHED OBSERVATORIES. 521 



2. Those which, placed at a A T ery elevated point, can only be visited 

 occasionally, and which must transmit their observations immediately 

 and regularly to the main observatory. 



Finally, we will speak of observatories suspended to captive balloons, 

 and which belong - to one or the other of the preceding categories accord- 

 ing to whether their observations require to be transmitted or not. 



Whatever arrangement be adopted, in order that it render service to 

 meteorology it should satisfy a great many strict conditions, the principal 

 of which are the following : 



1. The instruments should give exact indications. We do not speak 

 here, however, of excessive refinement, for meteorologists are agreed in 

 acknowledging that a difference of -^ or i of a degree Fahrenheit for 

 thermometric observations, or of gig or ^o °f an * neu in the height of 

 the barometric column, is not worth regarding. Readings to i degree 

 Fahrenheit for thermometric observations, andj^ or T l^ of an inch for 

 barometric height, are sufficiently accurate, the essential point being 

 that the records be always correct. 



2. The instruments should be as simple as possible, in order that the 

 chances of derangement be reduced to a minimum. 



3. The first cost should not be too great. 



4. It is important that human aid be dispensed with as much as pos- 

 sible ; the instruments should work and the records be made without 

 the assistance of the meteorologist. 



5. The registering apparatus should be so arranged that all the obser- 

 vations be collected in one table and recorded in the form of curves, so 

 that the movements of the instruments maybe seen at a glance. 



0. In the case of electrical connection between the detached and the 

 main observatory the number of electric conductors should be as small 

 as possible. This condition applies particularly in the case of a captive 

 balloon, since it is necessary to avoid increasing too far the weight of the 

 cable to be carried by the balloon. 



7. When a galvanic current is employed, the instruments themselves 

 should not form part of the circuit, because the working of the instru- 

 ments would be affected thereby, especially under the influence of the 

 sparks which pass at the instant of opening or closing the current; these 

 sparks, too, passing between metallic surfaces, would produce oxidation, 

 and consequently deterioration, of the instruments. For this reason the 

 use of mercurial barometers and thermometers furnished with platinum 

 wires hermetically sealed in the glass appears to me absolutely out of 

 the question. 



We do not need to decide on the form of barometer to be used, whether 

 the siphon or the cistern barometer, the balanced barometer employed 

 in Secchi's meteorograph, or an aneroid barometer. The method of re- 

 cording-observations which I have adopted allows of the use of all forms, 

 since even in the siphon barometer motion may be communicated to a 

 lever by means of a float, as in the dial barometer. 



