266 CARNEGIE INSTITUTION OF WASHINGTON. 



sailing vessel, where the sails will always interfere with an uninterrupted 

 view of the horizon at height of 18 feet above water. Heights much less than 

 18 feet are not desirable in investigations of the refraction. 



Among the mstruments available for measuring the dip-of -horizon at sea 

 are prismatic and reflecting circles, sextants, Troughton's dip sector, the 

 BHsh attachment, and Kohlschiitter's device for sextants, and the dip-of- 

 horizon measurer by Pulfrich, The prismatic circle, sextant, and dip-of- 

 horizon measurer have been tried on the Galilee first, subsequently on the 

 Carnegie, but none of these instruments has been found entirely satisfactory. 

 The prismatic circle was acquired primarily for land observations and was 

 found to be so unwieldly at sea that only a few experiments were attempted. 

 The principal objections to using sextants are the restriction imposed by the 

 limits of their arcs, the amount of computing usually required to obtain the 

 dip, and the periodic errors of the instrument, which are difficult to control; 

 these errors, of httle importance in navigation, are likely to be quite serious 

 in attempts to measure fluctuations in refraction no larger than a fraction 

 of a minute. The most promising instrument at first seemed to be Pulfrich's 

 dip-of-horizon measurer * made by Zeiss, of Jena. It consists of a low-power 

 telescope, at the object-end of which is attached a box containing a system 

 of three reflecting prisms. Rays from opposite points of the horizon enter 

 through perforations or windows, one on either side of the box. They are 

 reflected by the three prisms into the telescope. The images of the two 

 opposite portions of the horizon appear rotated 90° to a vertical position 

 with a light band, the sky, between them. The dip-of-horizon is one-half 

 the angular width of this band, provided, of course, that the instrument is 

 in perfect adjustment. The difficulties of observing with this instrument at 

 sea are increased by the motions of the images as the observer tries to hold 

 the instrument steady. The chief objection is a scissors-like motion which 

 apparently no amount of practice will eliminate; it is caused by changes in 

 the inclination of the telescope to the horizontal. 



Besides the obstacles to precision which are inherent in this instrument, 

 there are others which affect all instruments. The actual conditions of 

 visibihty, etc., of course, are taken care of by the observer's remarks, and 

 these observations may be weighted or rejected, as desired. One constant 

 source of error is in the height of instrument adopted. Did no refraction exist, 

 the dip-of-the-horizon could be computed for a given height of instrument 

 with more precision than it could be measured. The difference between this 

 computed dip and the dip observed being regarded as the refraction for the 

 given height, it will be in error by the same amount as the computed dip, if 

 the latter is in error. Suppose the given height of about 18 feet to be in 

 error by 1 foot, the refraction deduced will have an error of 0'.12. 



The height of instrument is measured in smooth water, as it is not practical 

 to measure it at sea, and it changes with the draft of the vessel, with her 

 rolhng and pitching and as she rides the waves and drops into the troughs. 

 Changes in draft are caused by the consumption of water, fuel, and stores, 

 and may be allowed for by assuming a linear change from port to port, where 

 the draft can be accurately read. The Carnegie's draft changes about 1.5 

 feet during a long passage. As to the ship's rolling and pitching, it is most 

 probable that observations for dip of horizon are never made on the Carnegie 

 when she is inclined as much as 10° to the vertical, since care is taken to 

 make them only on even keel as nearly as can be determined. This precaution 



*Pulfrich, C. : Ueber einen Apparat zur Messung der Kimmtiefe, Zs. Instrumentenk., Berlin, 

 1904, Heft 8, pp. 225-229. Moll, E. : Der Pulfrich'sche Kimmtiefeumesser, Hansa, Hamburg, 

 1906. 



