204 ANNUAL REPORT SMITHSONIAN INSTITUTION, 19 39 



is there. Such sensitivity might appear excessive for industrial 

 purposes, where ability to detect a trace of almost any element in 

 any given sample would seem valueless if nothing could be done 

 about its presence. Fortunately, something usually can be done. 

 Though in many cases the properties of metals are greatly influenced 

 by traces of impurities too minute to be eliminated conveniently by 

 ordinary methods of purification, addition of minute quantities of 

 other impurities will often offset undesired characteristics. In the 

 case of fusible alloys for safety valves or fire sprinklers, for example, 

 certain impurities if present by as much as 1 part in 10,000 will 

 raise the melting point by undesirable amounts. A relatively small 

 number of atoms in the crystal boundaries can produce a marked 

 effect, but atoms of other materials which will depress the melting 

 point again to its proper value can be added. The spectroscope is 

 invaluable in determining which impurities are responsible for such 

 effects, and which elements must be added to offset them. 



The outstanding value of the spectrograph comes in qualitative 

 analysis. Let us watch two scientists looking at a tiny piece of 

 lamp filament. To all outward appearances it is exactly the same 

 as any other piece of filament, but in an incandescent lamp it was 

 found to last twice as long as did filaments made from ordinary 

 wire. There is not enough material to analyze chemically, for the 

 sample is small and no advance hint tells what to look for. A 

 chemist using ordinary methods would be forced to guess as to 

 which elements were most probably present, and eliminate them 

 one after another. Long before he was finished such a small sample 

 would probably be exhausted. But here is a spectrograph; the 

 operator loads it with a photographic plate, opens the shutter, and 

 inserts a tiny piece of the filament into an electric arc placed in front 

 of the instrument. There is a flash of colored light and the spectrum 

 is recorded. Shifting the plate the operator repeats the process, the 

 second time burning a piece of an ordinary filament. When he 

 develops the spectrogram the story stands completely revealed. Each 

 type of atom in each filament has sent out its own group of waves 

 of different lengths, and produced its own pattern of spectrum lines. 

 By looking for differences in these patterns, and identifying the 

 elements from which the various lines which differ are known to 

 originate, the spectroscopist can see quickly and accurately which 

 elements are present in one filament and not in the other. 



The pattern of lines on a spectrum photograph may look compli- 

 cated and meaningless to the layman, but to the experienced spec- 

 troscopist it often tells a story as definite as that told by a line of 

 printing. This was perhaps not realized by the advertising agents 

 of a manufacturer of motorcars who recently strove to impress the 



