i 6 influence of a magnetic field upon the spark spectra of iron and titanium. 



4. Photographic Methods. 



The requirements as to photographic plates in an investigation of this sort are to some extent con- 

 flicting. Speed, a fairly tine grain, good latitude, so that weak lines may be obtained without serious 

 over-exposure of the stronger ones, together with enough contrast to give sharply defined lines, are ele- 

 ments not easily combined in one plate. A number of plates have been tried, including the Lumiere 

 "Sigma," the Seed "Gilt Edge 27," "23," and "Process," the Cramer "Crown" and "Inst. Isochro- 

 matic." Each kind of plate will give superior effects for a certain t^-pe of line; but in general I have 

 obtained the best results for the work from the Seed "Gilt Edge 27" for the blue end of the spectrum as 

 far as about X4600, and from there on into the red from the same plate bathed with the solution of pina- 

 cyanol, pinaverdol, and homocol recommended by Wallace (55). This plate is the best adapted of those 

 I have tried in regard to doing justice to all classes of hnes. It is a fast plate without an objectionably 

 coarse grain. The latitude is good. In the case of hnes of complex Zeeman separation, a plate with more 

 contrast will often fail to show weak components very close to stronger ones. 



A properly chosen developer will sharpen the hnes to a great extent, avoiding troublesome shad- 

 ing off from the central maximum. After trying several solutions, I have preferred a hydroquinone 

 developer giving strong contrast, due to Mr. Wallace, but not pubhshed so far as I know. The propor- 

 tions are as follows, using equal parts of A and B : 



Solution A: Solution B: 



Water. 48 oz. Water 48 oz. 



Hydroquinone 640 grains Carbonate soda (anhydrous) i oz. 



Sulphite soda (anhydrous) i oz. Carbonate potassium (anhydrous) 4 oz. 



Sulphuric acid (cone.) 30 drops Bromide potassium >^ oz. 



This developer does not stain the plates, even when warm. Development was usually carried to the 

 point where chemical fog sets in. This comes on slowly, and the solution is as efficient in bringing up 

 weak images as any I have tried. When used at 20 C a bathed plate is usually fully developed in 6 to 

 7 minutes. Some very good photographs were obtained for the region X5200 to X 5500 by the use of the 

 Cramer "Inst. Isochromatic"; but it was found best to soften its contrast by the use of a metol-hydro- 

 quinone developer. For the region X4800 to X5100, where the "Isochromatic" is weak, as well as for 

 the whole of the orange and red, the action of the bathed "27" has been unsurpassed by any plate used 

 in these experiments. 



5. Measurement of Magnetic Field. 



The accurate measurement of field-strength presented some difficulties in the case of iron on account 

 of the use of metallic terminals for the spark. The field for titanium was more easily obtained, and was 

 based on direct measurements by a bismuth spiral. This instrument was obtained from Hartmann and 

 Braun, but instead of using the regular formula for temperature correction, the spiral was sent to the 

 National Bureau of Standards and there calibrated to provide a series of curves for the variation of field- 

 strength with change of resistance for temperatures of 15, 20, 25, 30, and 35 C. When used at inter- 

 mediate temperatures the interpolation was simple. The resistance in and out of the field was measured 

 with a Kohlrausch bridge. 



A set of plates of the titanium spectrum, extending over the whole region investigated, was taken 

 with the magnetic field as nearly the same as possible. All parts of the magnet were left unchanged 

 and the same current was used throughout. By check measurements with the bismuth spiral and by 

 comparison of plates which overlapped enough to measure some of the same lines on both, it appeared 

 that a field-strength of 17,500 gausses was maintained for this set with a variation no greater than 200 

 gausses. Other photographs taken to supplement the measurement of certain regions had their values 

 reduced to correspond to a field-strength of 17,500 by comparison of the separationsof sharply defined lines. 



For the iron spectrum it is well known that indirect methods must be used to determine the field- 

 strength, since the use of iron spark terminals distorts the field to such an extent that any object as large 



