Grayson's Tin-land Plate. By A. A. C. E. Merlin. 7 



disturbance caused by the intervening negative lens can be com- 

 pensated. With the -jJo objective, amplifier and 6-micrometer 

 eye-piece, 131S divisions of the drum equal TQ Vo * n " > tnus tue 

 movement of the " wire " through one drum division represents an 

 interval amounting to y-yxsooo m - This ma y seem a P ract ically 

 unattainable degree of accuracy, but it must be borne in mind 

 that separating power is not here in question, and, to myself at 

 least, the wonderfully close agreement of the means of the first 

 five columns of measurements annexed hereto are sufficiently 

 significant, considering that they were effected with an objective, 

 the utmost separating limit of which could not exceed 94&00 ^ n - 



With reference to the annexed results, taking 763" 4, the mean 

 of the first band, as a standard (the mean of nine out of ten divisions 

 of another equally spaced (irayson plate in my possession is 763*6, 

 the first space of the band being rejected as obviously faulty, it 

 measuring only 75:2 -5 divisions), we find that the second column, 

 expressed in similar parts of an inch, varies from it by just under 

 one drum division, or 7^3^50 ^ n > tne means °f the second and 

 third columns agreeing within the surprising amount of T l <j of a 

 division, or reol M „ in.! The mean of column four exceeds the 

 standard by three divisions, while column nve shows an excess 

 of nearly two. It will lie noticed that fairly considerable dif- 

 ferences exist in the spacing of the individual lines in all the 

 bands, but I venture to submit that the accuracy of the measure- 

 ments is proved by the remarkably close agreement of the means 

 of the various columns, expressed, for convenience of comparison, 

 in similar and equal terms. So far as I can judge, the theory. of 

 probabilities renders it practically certain that such a close agree- 

 ment of five means can be due to no fortuitous coincidence. 



The sixth column exhibits the greatest variation from the 

 standard, falling short by the very considerable amount of 16*3 

 divisions, representing a difference of about 4^54 in. The 

 seventh column falls short of the standard by 7 ■ 7 divisions, and 

 the eighth by 9'S. Columns nine and ten show that the two 

 finest bands are wonderfully evenly ruled, their means being, 

 respectively, 4-8 and 8-4 divisions more than the standard. 



Since writing the foregoing, it has been thought desirable 

 that measured readings of two lines together throughout the 

 second band, three lines together throughout the third band, four 

 through the fourth, and so on up to ten through the tenth, should 

 be effected with the exact optical arrangement and magnification 

 employed for the original measurements. In this manner the 

 value of the whole ten bands is indicated in terms of the first, and 

 exactly the same portion of the screw is utilised for all, thus error 

 from differences in varying parts of the screw is eliminated, and 

 consequently the means of the five necessary readings in each of 



