Dip-Needle Errors 



365 



later), these larger undulations are quite similar in form. The range of inclination 

 covered only a part of such an undulation in the curve for No. 6; the undulation in the 

 curve for No. 2 is characteristic and unmistakably defined; that in the curve for No. 5 

 is of larger amplitude and extends over a somewhat wider range of inclination than those 

 in the curves for the other two needles. Each of these, again excepting that in the 

 curve of No. 1, first rises to a positive maximum as inclination increases and later falls 

 to a corresponding negative value, the ratio of the amplitude to the range of inclination 

 through which this feature extends being roughly the same in all the cases. 



In the curve for needle No. 1, between the values of inclination +56 and +59.5, 

 the order of minimum and maximum seems to be reversed, the minimum in this case 

 preceding the maximum, and the figure lacks the symmetry of form suggested by the 

 curves for Nos. 2, 5, and 6. The values given by needle No. 1 had to be rejected within 

 this range of inclination, being too discordant to be used in the mean of the other needles. 

 Several attempts were made to alter the grouping of the stations within these limits so 

 as to make it possible to draw a smooth curve. No satisfactory grouping was found. 

 An examination of the original observations disclosed that a radical change in polarity- 

 difference of needle No. 1 occurred some time between May 10 and May 17, 1916, 

 probably between the two sets of observations on May 14. Table 46, giving chrono- 

 logical groupings usually for three stations, shows how this change affected the relation 

 of the observed results with polarity A and polarity B. 



Table 46. Tabulation of Observed Data during January 1 to July 

 3, 1916, with Needle No. 1 in Dip Circle No. 177. 



Date 



1916 

 Jan. 1, 4, 8 

 Jan. 13, 17, 25 



Jan. 26, 30 



Feb. 2 



Feb. 6, 10, 13 



Feb. 25 



Mar. 2,17.... 

 Mar. 27,30... 



April 3 



April 6, 11, 24 



April 27 



May 2, 5 



May 10, 14 (1) 

 May 14 (2) ... . 

 May 17, 18, 21 

 May 24, 27, 30 

 June 1. 2, 7 



June 16, 28 



July 1, 3... 



Adopted in- 

 clination, / 



+ 58 58 

 + 56 2S 



09 



ir. 



58 55 

 + 58 23 

 + 58 39 

 + 58 30 

 + 58 07 

 + 56 45 

 + 56 51 



(I -I A) 



-16.6 

 -17.0 



-19.3 



-20.2 



-21.8 



-21.9 

 -19.0 



-17.8 



-15.8 



+ 7.5 



+ 6.2 



+ 4.6 



+ 0.8 



- 8.8 



- 9.7 



a -iB) 



+ 16.0 

 + 18.9 



+ 19.1 



+ 19.0 



+ 21.1 



+ 21.7 



+ 19.8 



+ 14.9 



+ 10.3 

 - 1.6 

 + 2.7 



+ 2.8 



3.6 

 6.6 

 3.4 



(A-B) 



+ 32.6 

 +35.9 



+38.4 



+39.2 



+ 42.9 



+ 43.6 

 +38.8 

 +32.7 



+ 26.1 



- 9.1 



- 3.5 



- 1.8 

 + 2.8 

 + 15.4 

 + 13.1 



Needle No. 1 was not used after July 3, 1916. The cause of this sudden change 

 in the behavior of this needle is not clear. The range in the inclination concerned is 

 relatively small, so that any change in balance due to a physical change in the blade would 

 be sensibly constant for all the observations after the date of occurrence, while any acci- 

 dent to the pivot would affect only the results of one polarity of the needle. Possibly 

 there may have been some combination of both of these conditions. Treating the obser- 

 vations before and after May 14 separately, the resulting correction-curves for needle 

 No. 1 are shown in Figure 7 and indicate clearly a time-change. 



Referring again to the short-period variations on correction-curves, it is quite certain 

 that those of larger magnitude are not caused by accidental observational errors. Such 

 results are obviously attributable to minute defects of pivots produced by corrosion, 



