Miscella7iies. 177 



8h. 15m. » . . - 31' 54.08" 



" " - - - - " 53.20" 



" " - ... « 55.19' 



" 20m. - - . . " 55.56" 



" " ■ - - - - " 55.64" 



" " . . - , " 55.00' 



" " - . - „ " 54.82" 



" 27m. - - . . " 54.95' 



the mean of which is 31' 54.81". It should, however, be remarked, that 

 a more perfect judgment can be formed of the exactness of contact of 

 sharply terminated points, such as were the cusps during the echpse, than 

 can be the case with edges or limbs, as tremulous as that of the sun, 

 where an alternate overlapping and recession leaves something to estima» 

 ; tion. On this account, an attempt to obtain several measures of greatest 

 distance of limbs was relinquished, both because greater inaccuracy was 

 apprehended from the above source, and the measures of the cusps af- 

 forded a more advantageous method of arriving at the same results. 



The maximum distance of the cusps, which may be obtained by inter- 

 polation from those nearest in point of time, will give the observe'd diame- 

 ter of the moon, free, it is believed, from the effects of irradiation. The 

 minimum distance will be a greatly magnified measure of the error of 

 the moon's assumed latitude, the ratio of increase of the distance of the 

 cusps at that point to the corresponding difference of the latitude being 

 about as 25.7 : 1.* 



At 5h. 20m. the measures of distance were relinquished, as the sun's 

 proximity to the horizon would soon render any further observation of 

 this kind of little value. 



At the end of the eclipse, the sun was scarcely 3 degrees above the 

 horizon, and the extreme undulation of his limb rendered much accuracy 

 in the time of the observation impossible ; and being, therefore, deemed 

 of little importance, it was not carefully noted, and may possibly be in 

 error. 



The sidereal clock, from which the above determinations of time were 

 taken and reduced, had been compared frequently during the months of 

 August and September with transits of stars, and the deviations of the 

 transit instrument, the value of the divisions of its level, and the irregu- 

 larities of the clock's rate, carefully registered and applied. From a 

 comparison of these observations, it appears, that the error of observation 



* The moon's latitude, as assumed for the calculation of the eclipse in the 

 American Almanac, is by the observation of the nearest approach of cusps, 10.05" 

 too large; a determinationvvhich, if the calculated semidiamctcrs of the sun and 

 moon be correct, is in error by only one- nineteenth part of the error of obser- 

 vation. - ' 



Vol. XXXV.— No. 1. 23 



