344 



SCIENCE 



[N. S. Vol. XXXIX. No. 1001 



give too low a value outside the atmosphere. 

 An equally plausible, and equally fallac- 

 ious argument is the following: It is said 

 that the density of water decreases with in- 

 creasing temperature at the mean rate of 

 about .00041 per degree from 0° to 100°. 

 Hence its density at 4° should be 0.99836, 

 but observations at 4° prove that water is 

 actually denser at this temperature than 

 at 0°, therefore the supposed decreased 

 density at 100° is a delusion. 



SOLAB CONSTANT WORK OP THE SMITHSONIAN 

 ASTEOPHTSICAL OBSERVATORY 



The earlier years of the work of the 

 Astrophysical Observatory were devoted to 

 the improvement of the bolometer and the 

 rise of it for the determination of the posi- 

 tions of lines in the infra-red solar spec- 

 trum. About 1902 attention began to be 

 devoted to measurements of the solar con- 

 stant of radiation. We approached these 

 measurements with a very much better in- 

 strumental equipment than that which had 

 been Langley's in the Mount Whitney 

 expedition of 1881. Soon after the Astro- 

 physical Observatory was founded, about 

 the year 1890, Langley introduced the 

 automatic registration of the galvanometer 

 in connection with the spectro-bolometer, 

 and in the subsequent years the difficulties 

 connected with the use of the recording 

 spectro-bolometer were so far overcome that 

 the solar spectrum could be observed from 

 the extreme ultra-violet end of the spectrum 

 at about 0.3 /x to a wave-length of about 3 fi 

 in the infra-red with great ease and accu- 

 racy, in an interval of 8 minutes of time. 

 Drift of the galvanometer, which in Lang- 

 ley's expedition to Mount Whitney he has 

 told me often amounted to a meter a 

 minute on the scale, was now so far reduced 

 that a centimeter an hour would be unusual. 

 In fact the bolometer, despite its great 

 sensitiveness, is about as easy to use for this 



work as an ordinary thermometer is for 

 measuring the temperature of the air. 



Our first measurements of the sun's 

 radiation as a whole were made with the 

 Crova alcohol actinometer, and in order to 

 standardize this instrument we constructed 

 a modified Tyndall pyrheliometer consist- 

 ing of a copper box fiUed with mercury and 

 having a cylindric bulb thermometer in- 

 serted radially into the box. Owing to the 

 difficulty of keeping the small thread of 

 mercury at the proper point for reading 

 purposes in the Crova actinometer, we 

 found it more desirable to develop the 

 pyrheliometer for our purpose. Soon a 

 solid disk of copper with a radial hole large 

 enough to enclose the thermometer bulb 

 was substituted for the box filled with mer- 

 cury, the use of mercury being limited to 

 insuring a good heat connection between 

 the bulb of the thermometer and the copper 

 of the disk. Some of these copper disk 

 pyrheliometers are still in use on Mount 

 Wilson. About 1909, however, the further 

 improvement was introduced of using silver 

 in place of copper for the disk. A thin 

 steel lining is provided for the hole where 

 the thermometer is inserted, so as to pre- 

 vent the mercury from alloying with the 

 silver. In these silver disk instruments the 

 thermometer stem, which is introduced 

 radially in the disk, is bent outside the 

 chamber at right angles so as to point 

 towards the sun. The whole instrument is 

 mounted equatorially with a device for 

 moving it by hand to follow the sun from 

 moment to moment. These disk pyrheliom- 

 eters, either of copper or silver, have now 

 been in use since the year 1906 with great 

 satisfaction. Their constancy over long 

 periods of time leaves nothing to be desired, 

 and the accuracy of observation reaches a 

 small fraction of 1 per cent. 



As the disk pyrheliometer is a secondary 

 instrument, it was necessary to develop a 



