3 



muligens tilskrives Ujevnhed i Chronometrets Gang, men 

 saameget bliver tilbage som Fejl i Eøjderne, ;it jeg anser 

 det rigtigst a1 antage, at mine Højder. med et rundt Tal, 

 kunne være beheftede med en sandsynlig constant Fejl for 

 hver Station af + LO". 



Paa den Nøjagtighed, hvormed Observationsuliret an- 

 giver Greenwich Middeltid, har jei: søgt at t'aa et Maal ved 

 følgende Overslag: 



• altitude amounting respeetively to 4-8" and — : 8". Sonie 

 part of this error may perhaps he ascribed to wanl of 

 uniformity in the rate of the chronometer; hut even withthis 

 deduction, the remainder is. 1 fchink, as an actual error in 

 the altitudes, suffioiént to warrant assuming that my solar 

 altitudes may hi' affected by a -probable constant error at 

 each Station of + 10". 



i lf the precision with which the chronometer used 

 for aoting the observations indicates Greenwich mean 

 time, I have sought to lind a measure as ibllows: - 



Observationsuhrets Sammenligning med Haveduhret. 

 Der toges flere Sam men ligninger for og efter Højdeobser- 

 vationerne. Af disse huder jeg for 1877, mit Duplexuhr, 

 en sandsynlig Fejl af den anvendte Ohrforskjel af + 0/15 

 (elter 4 Sammenhgninger) og + 0/11 af 4 Sammenlignin- 

 ger for og 4 etter Observationerne. For 187S tinder jeg 

 for Sammenligningen mellem mit Lommechronometer og 

 Chronometer Reid saavel i Soen som i Havn. Middelfejlen 

 af en enkelt Sammenligning + 0/10. I Regelen gjordes 

 3 Sammenligninger, saaat Middelfejlen ved en Sammen- 

 ligning tor eller efter 'Højdeobservationerne kan _ sættes til 

 + 0/06 og af Mediet af hegge til + 0/04. .leg kahler i 

 det følgende' denne sandsynlige Fejl A og sætter med et 

 rundt Tal A = + 0/1. 



Naar Tidssignal skulde observeres, var Regelen den. 

 at Skibschefen, Capt. Wille. forst sammenlignede Ohserva- 

 tionsuhret, et Lommechronometer. der slog 0/4, med.Hoved- 

 chronometret. derpaa gik i Laud paa Telegrafkontoret og 

 efter Tilhagekomsten ombord atter tog en Uhrsammenlig- 

 ning. .leg antager. elter et Skjon. denne Operations Re- 

 sultat at have en sandsynlig Fejl af +0/1, som jeg kalder A. 



Paa Telegrafkontoret observerede Capt. Wille Tids- 

 signalerne efter Observationsuliret. Den sandsynlige Fejl 

 at Observationen af et enkelt 'Signal antager jeg at kunne 

 sætte til 4> 0/2. Da i Regelen neppe mere end 2 af de 

 3 Signaler kunne antages at blive godt observerede (ved de 

 2 sidste er man forberedt paa Secundet), sætter jeg ilen 

 sandsynlige Fejl af Resultatet af Observationen af Tids- 

 signalerne til + 0/15 (D$). Ved en Lejlighed. da vi begge 

 observerede Tidssignalerne, stemte vor Bestemmelse af Ho- 

 vedchronometrets Stand paa 0/1. 



Ved Signalets Afsendelse paa Observatoriet i Chri- 

 stiania kan den sandsynlige Fejl, efter Vidnesbyrd fra ved- 

 kommende Astronomer. sættes til 0/15 pr. Signal. + 0/10 

 pr. 2 Signaler {Di). 



Ligeledes sættes den sandsynlige Fejl af Normalpen- 

 delens beregnede Stand for Christiania Sfjernetid. corri- 

 geret efter efterfølgende Tidsbestemmelse, til + 0/1 (D 5 ). 



Den sandsynlige Fejl af den nedenfor antagne Tids- 

 forskjel mellem Christiania og Greenwich Observatoriers 

 Meridianer sættes til + 0/2 (A)- 



Ved Længdeberegningerne er forudsat en jevn Gang 

 hos Hovedchronometret mellem de Tidspunkter, da dets 



Comparison of the watch or chronometer selected for 

 the observation with the cHef time-keeiier. Several com- 

 parisons were made- before aud after the observations of 

 altitude. Now, for 1 S77 (my duplex watch). I tind a prob- 

 able error of the assumed difference of the. errors of the 

 time-)iieces (4 comparisons) of + 0.*15, and with 4 com- 

 parisops before and 4 after the observations. of + 0.<11. 

 For 1878, I tind the mean error of a single comparison 

 between my pocket-chronometer and the box-chrdnometer 

 by Reid. both at sea and in harliour. to have been + 0.*10. 

 The number of comparisons having.as a rule been three, 

 the mean error of one comparison before or one after a 

 series of altitudes may be put -at + 0.*06, and the nr 'an 

 error of two comparisons. one before and one after, at 

 + 0/04. In the sequel I shall call this probable error 

 Di. and assume D L = + 0/1. 



The time-signals were g8nerally observed as follows: — 

 Shortly before their arrival, the commander of the vessel. 

 Capt. Wille. compared a pocket-chronometer. beatine 0/4, 

 with our chief time-keeper. He then went on shore to the 

 telegraph-office, observed the signals, and. on his return to 

 the ship. again compared the respective time-pieces. The 

 probable error of these comparisons on board may. I think. 

 be estimated at + 0/1, which I shall call D>. 



At the telegraph-office Capt. Wille observed the time- 

 signals with the pocket-chronometer mentioned above. The 

 probable error of the observation of one signal I have put 

 at + 0/2. Now. as only 2 of the 3 signals, on an average. 

 will be accurately observed (for the 2 last the observer is 

 prepared to the secoml). I shall estimate the probable error 

 of the result of our observations of the time-signals at + 0/15 

 (Di). . (Jn one occasion. when both of us (myself and Capt. 

 Wille) observed the time-signals. our determination of the 

 error of the chief chronometer agreed within 0/1. 



According to the estimate of the astronomers of the 

 Christiania Observatory, the probable error of one signal 

 as given with the key at the Observatory may be put at 

 + 0/15, of two signals + 0/10 (A). 



The probable error of the computed error of the 

 standard clock on Christiania sidereal time. corrected from 

 later transits of stars, is put at + 0/1 (D :> ). 



The probable error of the difference in time, as given 

 below.' between the meridians of the Christiania and Green- 

 wich Observatories. is put at + 0/2 (A)-' 



For computations of longitude, the chief chronometer 

 is assumed to have had a uniform rate between the moments 



1* 



