APPARATUS AND METHODS. I 7 



as the bismuth spiral or an exploring coil for the ballistic method will not give true values for the field 

 to which the spark vapor is subjected. It may be that the iron vapor, when sufficiently dense, has an 

 appreciable permeability of its own. There is, however, no e\ddence on this point. 



The plates for the iron spectrum were taken at intervals extending over a year, during which various 

 changes were made in the experimental arrangements which involved changes in the magnetic field. 

 However, a considerable region in the blue and violet was photographed with the same field, and the pub- 

 lication of Mrs. van Bilderbeek (49) gave an opportunity to make a comparison with her values. In her 

 work some photographs were taken using a spark with one iron and one zinc terminal, thus obtaining the 

 zinc triplet X 4680.3 17, as well as some iron lines in that region. Weiss and Cotton (20) by a series of very 

 careful measurements obtained the relation iAX/HX-'= 1.875 ^ 1 '* for the separation of the outer compo- 

 nents of this triplet, from which Mrs. van Bilderbeek deduced the value 32,040 gausses for the standard 

 field which she used when iron terminals alone were employed. I was able to select from my fist 33 lines 

 between the limits X3700 and X4400, which are also given in Mrs. van Bilderbeek's table, in nearly all 

 cases clear triplets, for which my measurements are of high weight. The ratio between Mrs. van Bilder- 

 beek's values and mine for these fines was in every case verj^ close to 2, the greatest deviation being given 

 by the value 2.14. The mean ratio for the lines is 2.01, giving a value of 15,940 gausses for the field used 

 by me in photographing the iron spectrum. This is in very satisfactory agreement with a value which 

 I had already determined by photographing the strong fine X 4383.720 as given by a spark between car- 

 bon terminals on which a little iron solution was placed in a field measured by the bismuth spiral as 

 17,600, and comparing the separation with that of the components of the same line very sharply 

 photographed with iron terminals used in the standard field. Exactly the same \-alue was given by com- 

 paring the separation of X4383.720 in two photographs, one with iron poles, the other in wliich the line 

 came up as an impurity in a titanium photograph taken with the standard titanium field of 17,500. 

 Assuming that the value of the field for iron was estabhshed by the other measurements, this last test 

 gave an excellent check on the standard field for titanium, which would otherwise depend on the meas- 

 urement with the bismuth spiral. It would seem then that the value of 16,000 gausses can be safely 

 taken for the standard iron field with an error less than i per cent. A considerable number of photo- 

 graphs for both iron and titanium were made with fields close to 20,000 gausses, sometimes slightly 

 higher, but the measurements were reduced to correspond to fields of 16,000 and 17,500, respectively. 



A similar system of checking field-strengths was appfied for the region to the red of X4400. A spark 

 was used with one terminal of iron and the other of brass. Two photographs were taken in which the 

 zinc triplet X 4680.3 17 appeared as well as a number of iron lines, among them the wide and sharp 

 triplet X4878.407. Using the value of Weiss and Cotton, the field-strength for the measured separation 

 of this iron fine (20,360 gausses for AX=i.389 A) was deduced. Spark terminals of the same kind with 

 all parts of the magnet unchanged were then used for a series of photographs covering the iron spec- 

 trum as far as X6700. The field was thus kept as nearly constant as possible, and by comparing the 

 separations of iron lines with this known field with those on former plates taken with various fields, it was 

 possible to reduce all values for the iron spectrum to the standard field of 16,000. 



6. Methods of Measurement and Reduction. 



The measurement of the earlier plates was carried out by Miss Wickham, while the later plates were 

 measured by Miss Griffin. The machine used was a small Gaertner comparator having a range of 8 cm, 

 the divided head reading to o.ooi mm. The process of measurement included the identification of Unes, 

 the determination of the reduction factor for the portion of the plate under examination and the measure- 

 ment of the separation of the Zeeman components. 



Various tables were used in the identification of lines. For the iron spectrum the tables of Kayser 

 and Runge (56) for the iron arc were supplemented by those of Exner and Haschek (57) for the spark. 



