DEPARTMENT OF TERRESTRIAL MAGNETISM. 265 



will be affected by the error of that value of the refraction which is used in 

 calculating the dip of the horizon. Although the atmospheric refraction is 

 usually small, reports of extraordinary values from time to time have shown 

 the desirability of extensive investigation, and a number of observations 

 have been made, principally by German investigators during the last 30 

 years. Koss, who prepared the first tables of dip-of-horizon that have a 

 temperature argmnent, has observed the horizon 10' above its normal position 

 and 3' below. A German epitome* states that the horizon has been observed 

 15' above and 3' below its normal position. Bowditchf says that rehable 

 observations have frequently placed it 10' above, and values as high as 32' 

 have been recorded. The significance of these figures will be realized when 

 it is remembered that each minute of abnormal refraction means an error 

 of one mile in the position of the ship. 



Such extraordinary values of the refraction at the horizon, as cited in the 

 previous paragraph, have not been found during 10 years of work at seas by the 

 Department of Terrestrial Magnetism, covering all the oceans. Whether large 

 abnormal values occur along the borders of equatorial and polar currents 

 shallows, and waters swept by very warm or very cold breezes blowing off 

 land, regions in which the cruises of the Galilee and Carnegie have not been 

 prolonged, requires further investigation. In all the observations taken first 

 on the Galilee, then continued on the Carnegie, amounting to 8,031 deter- 

 minations, the refraction has not raised the horizon more than 2'. 4, nor depressed 

 it more than 2'.0 below the position in which it would be seen if no refraction 

 existed. The observations were made mostly at heights above sea of 24 and 

 18 feet. The maximum raising of the horizon, +2'.4, was found on two 

 consecutive days, October 29 and 30, 1915, between New Zealand and Aus- 

 tralia; there was a heavy sea at the time and the horizon was noted as "rough." 

 The next maximum raising, +2'. 2, was also observed on two consecutive 

 days, February 9 and 10, 1908, in latitude 41° S., longitude 111° W., when 

 a very high sea, with clear, well-defined horizon, was noted. Values of 

 +2'.0 occur quite frequently, that is, about 1' above the average refraction 

 at sea. The maximum depression, +2'.0 occurs only once in the 10 years, 

 and was observed August 11, 1914, in latitude 71° N., longitude 5° W., 

 with well-defined horizon and good conditions generally; the position is on 

 the northern edge of the Gulf Stream, and not far from the ice floes of Green- 

 land Sea, in a region where atmospheric conditions are subject to marked 

 changes. Negative values are more rare than positive. 



All methods in use, up to the present time, for measuring the refraction at 

 the horizon, depend on one common principle — that of measuring the vertical 

 angle between two diametrically opposite points of the horizon or of measuring 

 the dijEference between this angle and 180°. This principle, common to all 

 methods, is objectionable because it must be assumed in general that the 

 refraction is the same for opposite points of the horizon. It is possible that 

 unusual values might be observed on occasions when the direction of the 

 abnormal value might be assigned with some degree of plausibility — for 

 example, when it is observed from near the edge of an ocean current that 

 may be plainly traced from the vessel, or when mirages occur limited to a 

 small part of the horizon, or by the adjustment of many observations taken 

 about the same time well distributed around the horizon. Experience, 

 however, on the Galilee and Carnegie, shows that the first two suggestions 

 could be used only on very rare occasions, and the last is not practical on a 



*Lehrbuch der Navigation, herausgegeben vom Reichs-Marine-Amt, second edition, Berlin 

 1906, p. 110. 



fBowditch, N.: American Practical Navigator, Washington, 1914, p. 117. 



