44 Bending of Magnetometer Deflexion-Bars, 
Bending experiments have now been made on magneto- 
meters by four makers. So far as elastic quality is con- 
cerned, all the bars from any one of the makers have been 
sufficiently alike to be grouped together. There have been 
differences, however, in the weights carried, which render it 
necessary to arrange the bars in seven groups when consider- 
ing the size of the bending effect. In every case, except that 
of a single special Dover bar, there have been at least two 
specimens in each group. What is here called a ‘ Cooke- 
Hlhott” bar is really an old Elliott bar adapted for use with 
Cooke magnets &c. in India. So far as elastic property is 
concerned, these may be classed with ordinary Elliott bars. 
The means of the results obtained appear in the following 

table. 
Increase of distance between magnets due 
to bending at distances stated. 
Maker or type of bar. KEX1OF® | EoclOe" (Unity = 1p = ‘001 nm., or 0001 em.) 
225 25 2625 30 35.40 45 50cms. 
| 
Cambridge Instrument Co.) 


379 | 110 are i a 31 36. 4h 45 ot 
me see See eae | om | 30  .... .85: . 42) ASG s eee 
pattern J newer......... ee ee ee 
Dover soi. {epeat’) 3% 8 fo a ee 
Eliott aul | 386. | eS a 
| } 
| . i 



The great majority of the measurements on which the 
calculations depend have been made by Mr. F. H. Smith, to 
whose skill and care as an observer the consistency in the 
results obtained in bars from the same maker owes much. 
Owing to the shape of the Cooke bars, the calculation of I 
in their case, and hence that of EH, is exposed to greater 
uncertainties than with the others. 
In actual magnetic work distances are measured only to 
‘001 cm., or to 10y, and to this degree of accuracy the results 
in the table are probably representative of the great majority 
of old magnetometers by the respective makers. 
In my previous paper I remarked on the fact that in many 
cases the increase in distance due to the bending varied 
roughly as the distance itself. This accidental phenomenon— 
accidental in so far as it is due to an undesigned relationship 
between the weight of the bar and that of the magnet and 
carriage—is shown by the bars of all the different makes, but 
