and combuclt'd Jii/ Piojessor Amici. 303 



tlie limb, is, that the object-glass can receive a greater quantit^ 

 of rays from one prism than from the other; so that we may by 

 this means render the images of two objects, which differ in 

 brightness, equally Imiiinous ; similar to what is obtained by 

 Hadley's reflecting sextant, by the elevation or depression of 

 the telescope upon the limb. But if by this motion we do 

 not obtain this equalitj' of light, on account of a too great 

 difference of brightness in the two objects, we apply to the 

 object-glass of the telescope the cover A (fig. 3.), half of the 

 circular opening of which remains uncovered, whilst the 

 other half is covered with a plane coloured glass. This 

 glass, being turned towards that prism which reflects the image 

 that is too luminous, will temper its too great brilliancy: 

 and, as the cover may be turned any way, this coloured glass 

 may always be placed before that part ol" the object-glass upon 

 which the prism throws too much light : thus, when we make 

 the observation with the instrument reversed, we can cause that 

 image to be reflected from the prism the brightest, which had 

 before been the least bright : and the mean of these two ob- 

 servations will then be exempt from the error of the parallelism 

 of the planes of this coloured glass, if such a defect exists. 



We may detect by this instrument the error of collimation 

 in three different ways. 



First. We may effect this by the coincidence or super-posi- 

 tion of two images of the same object ; the one direct, the other 

 reflected. The sun's disk is to be preferred ; but any ter- 

 restrial object whatever will answer this purpose, provided it 

 be not nearer than 50 toises; for it is only at this distance tliat 

 the parallax of the instrument begins to be imperceptible. 



Secondly. By the coincidence of two images of the same 

 object, exteriorly reflected from two sinall sides of the prisms. 

 In this case we shall have an angle of 90 degrees. Indeed the 

 two isosceles and rectangular prisms having then- great sides 

 jiarallel, when the vernier marks zero, will have turned 90 de- 

 grees, when the two smidler sides become parallel. 



Thirdly. By measuring two angular distances of two objects 

 almost diametrically oj)j)osed. The excess or defect of the 

 sum of these two angles uj)on 180 degrees will be equal to the 

 moiety of the angle to add to or deduct from the zero point 

 given by the vernier, in order to have the true zero, or the 

 error oi' collimation : — for example; Let the angles of two 

 objects be diumetiically opposed, the one of 85 and the other 

 97 degrees, the true zero, or the real beginning of the limb's 

 division, will be H5°4-97°= 18'^^— 180' = f°= 1". 



In comparing this last verification with that obtained by the 

 first method, ii there be a difference, it will be an error in the 



division 



