2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 65 



hand tube is withdrawn after the mineral grain has been centered, 

 so as to permit as much light as possible to pass through the spectro- 

 scope. A binocular microscope is not absolutely necessary, but 

 frequent readjustments of the scale and slit have to be made if the 

 mineral is observed by swinging out the upper part of the spectro- 

 scope and the slit holder. 



Light may be obtained from any source yielding a brilliant white 

 light, such as a Welsbach burner or a Nernst lamp, although sunlight 

 or daylight are objectionable because of showing the Fraunhofer 

 lines. For the study of minerals in thin sections, and in a few special 

 cases mentioned below, this is reflected up through the specimen 

 by means of the sub-stage mirror. In the majority of cases, however, 

 better results are obtained by concentrating the light laterally on the 

 specimen by a lens or by a parabolic mirror attached to the objective, 

 and observing the brightest portion of its path. Not only does the 

 latter plan yield the better spectra (apparently because they are 

 connected with fluorescence phenomena) , but it permits the examina- 

 tion of crystals on the matrix, gems in their settings, and other 

 similar objects, and, further, does not require any polishing or 

 special preparation of surfaces. The more intense the light the 

 smaller the grains which can be studied in this way. 



To set the wave-length scale of the instrument accurately a sodium 

 flame is used, scale division 058.9 1 being brought into coincidence 

 with the yellow (D) line. In addition, a small slip of " didymium " 

 glass, 2 which can be readily inserted at the opening where light for 

 the comparison spectrum enters, is very convenient, the interval 

 between the strong absorption bands of neodymium and praseodym- 

 ium in the yellow being set at about 058 (580 /x/x). See figure 1. 



The scale of the instrument is graduated in hundredths of microns, 

 but, except at the extreme red end, tenths of divisions can be readily 

 estimated, and it is most convenient to state measurements in three- 

 figure wave lengths. Since the edges of many of the absorption 

 bands are so hazy that they cannot be located exactly, and since the 

 positions of bands vary somewhat in different directions in aniso- 

 tropic substances,- as well as from one crystal to another in minerals 

 of variable composition, readings are liable to an uncertainty of 

 about 5 units. However, as the object of the present paper is not 

 to establish wave lengths, but to record the general characteristics 



1 This corresponds to wave length 589 \xn ; all measurements are stated in 

 the latter form. 



2 Obtainable from the Corning Glass Co., Corning, N. Y. 



