162 Transactions of the Society. 



between the rulings. By this plan setting errors are minimized, 

 and the distance from centre to centre of the rulings is shown. As 

 a matter of fact, there are occasional slight variations in the breadth 

 of the lines themselves, and thus measures merely from edge to 

 edge of a ruling may not indicate the exact distance from centre 

 to centre. In the older stage micrometers * the errors of spacing 

 were so gross that refinements of the kind were not usually 

 employed, or necessary, in their measurement, and indeed, we find 

 that the screw micrometer still almost universally used, with no 

 means of accurately adjusting its " fixed " wire, is a survival of 

 the rough methods now utterly inadequate when dealing with such 

 a masterpiece as Grayson's new plate. The Nelson-Powell screw- 

 setting micrometer is alone suitable for the purpose of all accurate 

 woidv, for the bi-filar kind, where each wire is independently 

 moved, introduces complicated vscrew errors. 



A glance at the table will make it evident that the means 

 exhibit little variation, for, taking 920 '0 as the standard, the 

 greatest divergence on each side does not amount to quite four 

 drum divisions, or 3j3o 1 ooo" m -» ana - mos t °f the means are much 

 closer. In the measurements of the individual spacings, there is 

 an apparently somewhat marked inequality in the first yo\jo i n - 

 space amounting to 6 • 3 divisions above the 920 ■ mean, but this 

 only represents a difference of about yyoWo m - 



In addition to standardizing the new plate, meaned measure- 

 ments were taken of the first ten spaces of the last six bands 

 using a Powell objective of I.M.P. 122 and 1*27 KA. This was 

 employed with a negative amplifier, it being furnished with a 

 correction collar, which enabled critical definition to be thus 

 obtained with a W.A. of 95. It was judged inadvisable to 

 measure the four coarser bands with this combination, as a y^-j in. 

 space would extend over more than three-fourths of the entire 

 field, making parallax any other errors possible, even when all 

 precautions are taken. The results are shown in the second set 

 of columns of the annexed table, the intervals representing the 

 distances from the centres of the lines. It will be seen that the 

 extremely minute inequalities of these bands are rendered very 

 manifest through the enormous magnification, which causes one 

 drum division to only equal about 2087O00 m -t 



The next step consisted in the measurement of the diameters 

 of the first two lines of each band, which proved quite feasible 

 with the optical arrangements above specified. The highest of the 

 20 readings thus obtained was 49 '6 drum divisions, the lowest 

 41*0, and the mean was found to be 43*7 drum divisions. 



With the magnification employed, 43 • 7 drum divisions represent 



* See ' Carpenter,' 8th ed. p. 275. 



t This is more particularly applicable to the complete table of measure- 

 ments preserved in the Society's Library. 



