November C, 1003.] 



SCIENCE. 



593 



tiule by lunar observations, and the ac- 

 enraey of the work has since been fully 

 confirmed. 



The observations for longitude were con- 

 tinued through the months of October, No- 

 vember and December, 1874. During this 

 time over 700 meridional transits of the 

 moon, thii'teen occultations of stars, and 

 about 540 zenith distances of the moon's 

 upper and lower limbs, combined with those 

 of well-known stars near the moon, were ob- 

 served. The first reduction of the observa- 

 tions apparently agreed with the result 

 obtained by Professor Lyman, viz., 10 h. 

 31 m. 15 s., but after returning to Europe 

 and correcting the tabular right ascension 

 by the contemporary observations made at 

 Washington, Greenwich, Paris, Konigs- 

 burg, Strasburg and the Cape of Good 

 Hope, Captain Tupman increased the result 

 by about twelve seconds of time. Ilis final 

 result by meridional transits of the moon 

 was 10 h. 31 m. 26.0 s. ; by zenith distances, 

 10 h. 31 m. 27.3 s.; by occultations of 

 stars, 10 h. 31 m. 26.9 s. 



The result officially communicated to the 

 Hawaiian Government Survey was 10 h. 

 31 m. 27.2 s., upon which all the maps 

 since then have been based. It is intei-est- 

 ing to note that this value agrees to the 

 tenth of a second witli the latest determina- 

 tion of the longitude. 



Captain Tupman, however, afterwards 

 weighted the above results according to the 

 number of observers employed on each, giv- 

 ing the occultation result the weight 5, the 

 mean of the fii-st two results the weight 4 

 and M. Fleuriais' result the weight 1, on 

 which conditions the resulting longitude is 

 10 h. 31 m. 26.3 s. If he had used the 

 value deduced from Fleuriais' work by the 

 A.ilroiiomixchc Xachrichtrn, his final mean 

 would have been 10 h. 31 m. 26.7 s. 



In Jfareh, 1875, Captain Tupman made 

 an attempt to connect Honolulu with San 



Francisco by transpoi'tation of chronom- 

 eters. Accordingly twenty chronometers 

 were carried by H. B. M. S. Reindeer, 

 Commander C. V. Anson, from Honolulu 

 to the U. S. Navy Yard, Mare Island, 

 and compared with the local time at 

 both stations. Unfortunately the Rein- 

 deer was blown out of her course by a 

 northerly gale, which lengthened her voy- 

 age seven or eight days, and lowered the 

 temperature in the chronometer boxes as 

 much as 15° F. Hence the resulting de- 

 termination of the longitude of Honolulu, 

 viz., 10 h. 31 m. 33.2 s. ± 3.0 s. w., had 

 very little value. 



Again, in August and September, 1884, 

 an attempt was made by the Hawaiian 

 Government Survey with the cooperation 

 of Professor Davidson and Mr. Morse in 

 San Francisco, to determine the longitude 

 of Honolulu by comparing the chronom- 

 eters on board the 0. S. S. Co. steamer 

 Mariposa, with the local time at each end. 

 This was done for two round trips, giving 

 a mean result of 10 h. 31 m. 25.8 s. 



In view of all the above facts, the 

 Hawaiian government adopted 10 h. 31 m. 

 26.0 s. as the most probable value, in rating 

 chronometers and furnishing standard 

 time, but not for mapping, until the recent 

 telegraphic determination made by the U. 

 S. Coast and Geodetic Survey. 



John F. H.wpord. 



SCTENTfFW BOOKS. 

 A Discussion of Variable Stars in the Cluster 

 w Gentauri. By Solon I. B.\ilev. An- 

 nals of the Astronomical Observatory of 

 Harvard College, Vol. 38. 4°. Cambridge, 

 Mass. 1902. Pp. 252; 13 plates. 

 Among the most interesting discoveries in 

 the subject of variable stars during the last 

 decade belongs the finding of an exceptionally 

 large number of variables in certain globular 

 star elustcrs. The remarkable fact that in 

 many of .these systems a not insignificant 

 proportion of all the stars change their light 



