TEMPERATCTKE OF THE SURFACE OF THE MOON. 27 



by sliding the lamp along the scale, bj' means of its cord, until these two squares of light were of 

 equal iiitonsity. Then, if the intensity of the standard is Icnown for all points of the scale, we 

 obtain the intensity of the suuliglit or nioonliglit at that part of the spectrum. If the wheel- 

 photometer was used, a proper factor must be introduced to give the degree of diminution caused 

 by it. Eight points in the spectrum at which comparisons were to be made, which we may roughly 

 designate by their aj)proximate colors, were selected. Their wave lengths, and the settings of the 

 grating circle, together with those for several of the Frauenhofer lines, are given in the annexed 

 table. 



A table giving the intensity of the illumination in the observing telescope, obtained from the 

 photometer lamp for each decimeter of the lamp-scale, was next constructed from data obtained 

 by observation. The assuni])tion which has been made in similar photometric measures that the 

 intensity of the illumination is inversely proportional to the square of the distance of the lamp- 

 flame from the slit, leads to results which may be considerably in error, particularly if some of the 

 observations were made when this distance was small. The reasons for this are various. Jf, starting 

 with the lamp at the end of its scale, we slide it gradually forward toward the slit, the intensity 

 of the light in the observing telescope will increase gradually until the aperture of the collimator 

 is filled, and then on closer approach the intensity no longer increases but remains constant, whereas 

 by the law of inverse squares it should increase from a certain value up to infinity at the slit. In 

 the apparatus used in these experiments this constancy of illnmination began at about 2 or 3 

 decimeters, and measurements made with a smaller scale reading than o decimeters were avoided 

 as much as possible. On account of the small proportion of blue and violet rays iu the lamp-light, 

 however, it was sometimes necessary to make the comparisons in the upper end of the spectrum 

 with the lamp so near the slit that the value of its light intensity was subject to considerable 

 uncertainty, and it is for this reason that the great difterence in quality between the standard and 

 the lights to be compared is so prejudicial to the accuracy of the observations. 



It was preferred, in making the measurements, to diminish as much as possible the violet of 

 the sunlight or moonlight by means of the wheel-photometer, thus enabling the comparison to be 

 made with the lamp at a greater distance from the photometer slit. 



The edges of the lampflame are considerably more brilliant than the central portions. When 

 the lamp is near the extremity of its scale, its effective brilliancy is the average of that of all its 

 parts, but when brought up close to the slit, the effective rays are those from the central portions 

 only. From both this and the foregoing reason, the decrease in the brilliancy of the light in the 

 observing telescope as the lamp is moved away from the slit is less rapid than that required by the 

 law of inverse squares. 



The law actualij- followed was determined empirically by means of the wheel-photometer. A 

 second kerosene lamp with Argand burner, quite similar to the standard lamp, was placed directly 

 in front of the slit, at such a distance that when matched by the standard lamp the scale reading 

 of the latter was a little less than 5 decimeters. Having determined this reading by taking the 

 mean of five settings, the wheel-photometer with its index set to 10 (or with its apertures open to 

 their full width), was interposed between the auxiliary lamp and the slit, cutting down the bril- 

 liancy of the direct light to one-half; and the new position of the photometer lamp, when matched 



