382 Mr. J. S. Ames on the Concave Grating 



kept of thermometer- and barometer-readings ; for the cor- 

 rections due to variations in temperature and pressure may 

 be considerable. 



Since no absorbing solution is known which lets through 

 the ultra-violet rajs alone^ the following method has to be 

 used to determine what lines on any negative are ultra-violet 

 ones. A compound negative, as just described, is taken, 

 having all the overlapping spectra at the point in question 

 along the middle of the plate, and the visible lines alone, 

 obtained by insez'ting absorbents, along the top and bottom. 

 Those lines present in the first and not in the second are then 

 ultra-violet ones. 



The time of exposure varies. For arc or solar light five 

 minutes is the average time required for the most sensitive 

 part, in the third spectrum on plates registering 18 on War- 

 nerke's sensitometer. Ten minutes are required above the 

 D lines in the second spectrum, using erythrosin plates. One 

 hour is needed for cyanine plates, photographing down to the 

 C line. As a practical example, the entire iron and solar 

 spectra were photographed in the second and third spectra 

 from the D Hnes down to the extreme ultra-violet in nine 

 hours. This includes time spent in developing. Thirty 

 plates, each 19 inches long, were exposed, giving of course 

 many duplicates. Only 10 plates are necessary in the second 

 spectrum of a 20,000 grating for the whole spectrum from the 

 D line to the extreme ultra-violet, wave-length 2000. In one 

 case Liveing and Dewar used 170 plates for the ultra-violet 

 spectrum alone. 



With a very powerful induction-coil, worked by a Siemens 

 alternating dynamo, with 6 gallon-Leyden jars, 10 minutes is 

 enough in the most sensitive part and 30 in the extreme ultra- 

 violet, wave-length 2200. 



A compound negative taken in the above manner is placed 

 on a dividing-engine, and measurements made on the lines of 

 the two spectra, using a low-power microscope with a single 

 stretched cross-hair. Since the solar spectrum continues 

 down to 3200, the same orders of the two spectra can be com- 

 pared thus far. Beyond this it is necessary to use different 

 orders. For instance, wave-length 2800 in the third spectrum 

 can be compared with solar lines about wave-length 4200 in 

 the second. This same method is used to determine the 

 relative wave-lengths of the solar spectrum. 



To enlarge photographs with a scale of w\ave-lengths, like 

 Rowland^s map of the spectrum, one must proceed as follows. 



To make the scale, a thick plate of glass, slightly longer 



