34 MEMOIKS OF THE NATIONAL ACADEMY OF SCIENCES. 



The adjustmeuts are coiuiuouly made so that heat falling ui)on the bolometer causes a deflec- 

 tiou of the image to the south, thus increasing the reading on the scale, whose zero is at the 

 northern end. It may be added, in further indication of the sensitiveness of the instrument, that 

 on bolometer No. 1 (whose resistance is 80.5 ohms) by Matthiessen's table, the change of tempera- 

 ture, corresponding to a change of resistance of 0.0001 ohm, is 0O.00032 C. Accordingly, when the 

 needle is in such a condition of sensitiveness that it executes a single vibration in 10 seconds, and 

 if we employ a current of 0.1 ampere, a change of one division on the scale corresponds to a change 

 of temperature in the bolometer strips of 0^.000016 C. This result is to be uiulerstood as merely 

 ajiproximate, and as indicating nearly the limit of sensitiveness attained in actual work at present. 

 It need hardly be added that greater nominal setisitiveness can be obtained to almost any extent 

 by increasing the time of swing: but the gain is apt to be only nominal, for we are to consider 

 that, other things being equal, the efliciency of the instrument increases as the probable error 

 diminishes, where this probable error is expressed as a fraction of the deviation in question. In 

 fact, as the concentrated moonbeam drives the image off the scale altogether in the above condition 

 of sensitiveness, it is necessary to employ the damping magnet, not to increase, but to diminish, 

 the time of vibration, so that the nnage may remain on the scale. Under these latter conditions 

 the probable error of a single observation is very small, i)robably not exceeding 2 per cent.* 



DESCRIPTION OF BOLOMETERS EMPLOYED. 



Bolometer No. 1, which has been chiefly used in measurements of total lunar radiation when 

 concentrated by the concave mirroi-, and for comparative observations with the Leslie cube, has 

 a square central aperture, 8.3 millimeters on a side, through which the blackened strips of the 

 central or exposed arm may be seen, presenting to the incoming rays an ai'ea of 49 square millime- 

 ters, and composed of 23 thin strips of blackened platinum, each about 0.001 millimeter thick, in two 

 tiers, the rear ones covering the apertures left between the front ones. The other, or protected 

 arm, is made up of 24 strips, an extra protected strip being introduced in the circuit of the exposed 

 arm, to balance the resistance, which is 80.5 ohms for either arm. The case is a cylinder of ebonite, 

 projecting so far beyond the strijis as to cut them off from all radiations, except those from the 

 subject of experiment. 



Bolometer No. 13 is composed of 8 side strips and 9 central ones, each 0.25 millimeter wide, 

 the latter forming a band 2.3 millimeters wide and 10.3 millimeters high, with which measures in 

 the lunar spectrum have chiefly been made. Each arm resists 38.4 ohms. 



Our direct observations on the lunar heat may be grouped under six divisions. (1) Quantita- 

 tive measurements of lunar heat as compared with solar; (2) comparisons of the moon's heat with 

 that from a terrestrial source; (3) the comparative transmissibility of our atmosphere for lunar 

 and solar heat; (4) comparative transmissibility of glass for lunar and solar heat; (5) heat ob- 

 servations during a lunar eclipse; (G) the formation of a lunar-heat spectrum. 



Class 1. — Quantitative measurements of lunar heat as compared with solar. 



Let US expose the bolometer to the lunar radiation, either direct or concentrated, and note 

 the i-esultant galvanometer deflection, and repeat the experiment the next day with the solar 

 radiation, diminished in a known ratio. If the moon be full and at an equal altitude with the sun 

 at the time of observation, we have the direct ratio of heat received from each at the earth's surface 

 but it is to be remarked that we cannot confine these observations to the single night of full moon 

 without giving inordinate time to the research (since they should be often repeated); while, if we 

 take them at times much before or after the full, considerable errors may be introduced by our 

 ignorance of the true law of the variation of the moon's heat with the phase. Where we have been 

 obliged to use the latter class of obsei'vations, we have reduced them by Zollner's law. It is to be 

 observed, also, that it is not only more than doubtful whether the transmissibility of the atmosphere 



* It appears in Lord Rosse's observations that the mean of a series of 10 gave a jjrobable error of 19 per cent, with 

 the thermoj>ilc and galvanometer then employed. Accordingly, if we do not consider constant errors, but only acci- 

 dentiil ones, we find that, owing to the increased sensitiveness and steadiness of our .ajiparatiis, a single meaaurement 

 with the present train is equivalent to several hundred ot that employed by Lord Eosse. 



