12 



INFLUENCE OF A MAGNETIC FIELD UPON THE SPARK SPECTRA OF IRON AND TITANIUM. 



Fig. 2. 



(b) A stronger light from the spark was obtained for observation along the axis by not insulating the 

 pole-tips, using one solid and the other pierced as in (a), with the spark between iron tips at the ends 

 of brass rods held vertically between the magnet poles by means of a wooden frame or the brass and fiber 

 holder described on p. lo. Titanium terminals were held in the simple clamp described above. This 

 worked well for getting the "longitudinal effect" (^-component) without the introduction of a Nicol 

 prism in the optical system. Such an end-on arrangement is of course necessary for the study of the 

 circular polarization of Zeeman components. However, for general work in measuring the separation 

 of components, this method has the disadvantage that there is a considerable increase of field-intensity 

 close to the magnet poles, amounting with some gaps to 25 per cent, as well as an inequality at the two 

 poles resulting from one of them being pierced, so that the sharpness of the Zeeman components is not 

 all that could be wished. 



(c) The most useful method, and that used (with varying shapes of the pole-tips) for almost all of 

 the best observations, was to set the magnet at right angles to the direction at which the Hght was 

 observed, use both pole-pieces solid, and separate the light by means of a Nicol prism over the sUt into that 

 vibrating in a plane at right angles to the magnetic force-lines, or parallel to these. This arrangement 

 made it possible to photograph successively the Zeeman components given respectively by vibrations 

 perpendicular and parallel to the force-lines by turning the Nicol prism through 90, leaving the magnet 

 unchanged. Furthermore, by projecting the image so that only the hght from that part of the spark 

 midway between the magnet pole-pieces falls upon the slit of the spectrograph, the 

 change of field near the pole-pieces does not disturb the definition of the Zeeman com- 

 ponents. Even if the sht is long enough so that parts of the image come from regions 

 of different field-strength, the spectrograph, not being astigmatic, shows merely a 

 wider separation toward the ends of the components, the sharpness not being affected, 

 so that accurate measurements may be made by selecting the narrowest portion of the 

 separation. 



Three forms of magnet pole-tips were used with this arrangement. In the first, 

 the conical tips ended in circular faces 6 mm in diameter. Tliis was used for most of 

 the work on iron and for the earlier work on titanium. With titanium, however, the 

 pieces of metal were irregular in shape and often rather large, so that with a short 

 magnetic gap it was difficult to bring the terminals close enough together to avoid sparking to the 

 magnet pole-pieces. The later and best set of titanium plates was taken with pole-pieces somewhat 

 chisel-shaped, made by miffing out opposite sides of a conical tip of 12 mm face to a thickness of 1.5 mm. 

 The thin ends were then placed parallel to each other and in a line with the beam of hght passing to the 

 slit. This gave a very uniform field for the hght of the thick spark, part of the vapor of which might 

 otherwise have gotten into weak portions of the field. Probably the best design is a modification of that 

 just described, in which the chisel edge was left 3 mm in thickness and 12 mm long, and not so deeply 

 milled as before. This form of tip gave a very strong field and a gap of 6 mm could be used without diffi- 

 culty. The drawing in Fig. 2 shows this design, with which a number of the later iron spectra were taken. 



A current of 10 to 12 amperes from a 1 2. 5 -kw generator was generally used for the magnet circuit. 

 15 amperes could be used for runs of two or three hours, but the magnet rapidly became heated. This 

 current was almost sufficient to saturate the core and a larger current gave but a small increase of field. 

 The heating of the core by long-continued runs, even at 10 amperes, was considerable in warm weather, 

 when the two electric fans used to blow the sparks, and which also played on the magnet, e.xerted httle 

 cooling effect. Almost at the close of this investigation a very efficient means of cooHng the core was 

 devised. Injuries to the insulation of the wire made it necessary to rewind both magnet coils. When 

 the cores were laid bare, a spiral of soft copper tubing of 6 mm outside diameter and 4 mm bore was 

 wound around each core next to the iron. Strips of "5000-volt hnen" were wound over the spiral as 



