MOUNT WILSON OBSERVATORY. 251 



ness of air than hitherto has been considered necessary. With a tungsten lamp 

 serving as the source of white hght, minimum thicknesses of 7 and 40 meters, 

 respectively, are found to give the A and B bands, while 9.5 meters of air of low 

 humidity show the water-vapor band a. 



CAUSES OF LYMAN GHOSTS IN DIFFRACTION GRATING SPECTRA. 



Mr. Anderson has made a study of the Lyman ghosts shown by three con- 

 cave gratings in regular use at the Observatory. These are the laboratory 

 4-inch grating of 1 meter radius, the 5-inch of 15 feet radius, and the 4-inch 

 of 1 meter radius in use on the mountain. The data thus obtained, together 

 with those already published by Lyman, and by Meggers and Kiess, have 

 been used in a discussion of the probable origin of this type of ghosts which 

 will appear in the Journal of the Optical Society of America. 



VACUUM THERMO-COUPLES. 



Laboratory work on the improvement of vacuum thermo-couples and the 

 application of them to astrophysical problems has been continued by Mr. 

 Pettit and Mr. Nicholson along the lines indicated in the last annual report. 



About 20 couples have been constructed for observational work on the 

 mountain, and several piles have been made for laboratory work. The couples 

 used are of the compensated type, with the two junctions brought as closely 

 together as possible. The strips employed are of bismuth and of bismuth 

 alloyed with 5 per cent of tin. The mass of the entire couple as now con- 

 structed, including the two tin receivers 0.5 mm. in diameter, is only about 

 0.025 mg. 



These couples are mounted in an evacuated cell of pyrex glass so constructed 

 that they may be removed without destroying the cell or couples. The 

 vacuum increases the sensitiveness of these couples to from 10 to 15 times its 

 value at atmospheric pressure. The effect of the vacuum is slight until a 

 pressure of about 1 mm. of mercury is reached, when the sensitiveness begins 

 to increase rapidly. 



A theoretical investigation of the thermal and electrical properties of these 

 couples has been made with the following results: (1) the time required for the 

 current produced by the thermo-couple to come to a maximum after the 

 radiation strikes the receiver is only a few thousandths of a second, and not at 

 all comparable to the period of the galvanometer; (2) the eJB&ciency of the 

 thermo-couple is very low, being in absolute numbers about 1 X 10""®, and rela- 

 tive to the perfect heat engine, about 1 X 10~^. From this it appears that if all 

 losses could be nullified, it would be possible to construct a thermo-couple 100 

 times as sensitive as those now used; (3) the loss by conduction of heat along 

 the strips is greater than the loss by radiation; (4) the gain obtained by placing 

 the thermo-couple at the center of curvature of a hemispherical mirror would be 

 comparatively slight. 



The galvanometers generally used with these thermo-couples are of the 

 D'Arsonval type, "Leeds and Northrup high sensitivity." With instruments 

 of from 10 to 60 ohms resistance, scale-distances of as much as 11 meters have 

 been employed. The sensitiveness of the 17-ohm instrument is 3X10~^° 

 amp./mm. and is regularly used in the stellar work at a scale-distance of 7 

 meters. With this arrangement a temperature difference of 0.000001° C. will 



