233 



The collimator in its perfect state consists of a piece of cast iron 

 8 inches long, 4 wide, and from ^ to \ an inch thick, having two up- 

 rights in the form of Y's, to which the collimating telescope is firmly 

 fastened. The support is then floated on mercury in a deal box, 

 somewhat larger than the flat portion of the iron, and having its bot- 

 tom just covered with mercury. The float is kept in its situation in 

 the middle of the box, and prevented from moving horizontally by 

 two smooth iron pins projecting from its sides, and moving freely in 

 vertical polished grooves of metal let into the sides of the box. The 

 whole of the telescope projects above the edges of the box, and a 

 screen of black pasteboard with an aperture equal to that of its ob- 

 ject-glass, is fixed to the end of the box to keep off false light. The 

 instrument was placed on a table attached to the wall of the observa- 

 tory, and directed (by looking through the telescope) to the wires of 

 a fine achromatic furnished with a wire micrometer. The cross wires 

 of the collimator were then illuminated by a small lantern placed 

 behind its eye-glass with oiled paper interposed. 



The object of the author in this arrangement being to ascertain 

 the limits of variability, in the position assumed by the collimator, it 

 was deranged purposely in a variety of ways, by removing and re- 

 placing the float, or carrying the whole instrument from its place, 

 and every method he could think of used that could fairly introduce 

 error. His preliminary trials were made with a wooden float ; but 

 this was soon laid aside after ascertaining that the greatest single er- 

 ror committed in using it, did not exceed 2"'5S in the position of the 

 horizontal point. Other floats were then tried, and it was found that 

 the increase of their length and browning their surfaces with nitric 

 acid produced material advantages. In 151 single results thus ex- 

 perimentally obtained, 28 only were found to give errors in the de- 

 termination of the horizontal point exceeding 1", and only two 

 amounting to 2". But if the means of every successive 5 be taken, 

 and the experiments with the wooden float rejected, the greatest er- 

 ror did not exceed 0"'4, and even here the influence of a constant 

 source of error depending on the support of the micrometer employed 

 was apparent. 



The author then describes at length the mode of using the colli- 

 mator and of observing with it. The instrument hitherto described 

 may be called the horizontal collimator, but he then proceeds to de- 

 scribe a vertical collimator, in which the telescope is fixed perpendi- 

 cularly to the float and placed immediately under the axis of the 

 circle. By this arrangement the necessity of transporting it from one 

 side of the observatory to the other is avoided, the reverse observa- 

 tion being made by merely turning the float half round in azimuth. 



It is not necessary that the telescope of the collimator should have 

 a tube, nor does the author appear to regard its length as of any im- 

 portance, it being merely the direction of its axis which is the subject 

 of examination ; and the accuracy of this examination will depend on 

 the length and power of the telescope of the circle to be collimatcd. 



