<<

ASTRONOMISCHE NACHRICHTEN. 3229. Band 135. No .13.

~ ~ ~ ~ ~~~

The Photometric Catalogues of the Harvard College Observatory. By Adward C. Piekering.

Mr. S. C. Chandler, formerly employed in the Harvard comparison will reveal. Such imperfections are inevitable I College Observatory, states, in a recent number of the in such work. Astronornische Nachrichten (Vol. I 34, p. 35 j), that there are Fifteen instances of error are pointed out in the total certain errors in the photometric catalogues of that Obser- of eighty- six variable whose observations are printed vatory, and draws certain conclusions as to the general I in Volume XXIV, and we are given to understand that value of the whole work. As shown below in detd the i similar errors probably exist throughout the entire catalogue particular errors specified are in general confirmed, but the of over twenty thousand stars contained in that volume. inferences he has drawn regarding the whole work are not I Every astronomer is of coiirse aware that the difficulties justified. Whatever animus underlies the attack can only of identification in the case of variable stars, especially be important in so far as, if it exists, it tends to weaken ~ when they are faint, are much greater than with stars the force of sweeping assertions and deductions sought to I photometrically constant. It is somewhat as though it be drawn from scanty premises. Animus unfortunately is i shoiild be argued from a physician’s losing twenty per cent not unknown in scientific discussion, and may exist, as we I of his cholrra patients thdt he had been equally unfortunate believe it exists, if at ail, in this case, without the person in his general practice. whom it dominates being aware of its presence. ! No evidence is afforded by the published criticism ‘I‘he real scientific intcrest lies in the yuestions ns to I that the writer has seen the statement of the precautions how far the results of the work as a whole are affected in taken in the matter of identification contained in the Annals, I value for the purposes of those who are to use them, and Vol. XIV, p. 6, and Vol. XXIII, p. 4. He states that the whether better methods than those pursued are attainable. I process involves, if he understands it aright, the turning of Let us take tirst the stars of Volume XIV. One error only the mirrors reflecting the into the field of view rfor- is pointed out, narncly that as to BJ). +z~’?~zs.This had I ward or back in some instances several degrees from the been detected here and an explanation forwardcd some meridian in search for the desired objectsa. ! months ago to the Suptrintendent of the British Nautical ‘l‘he di.imeter of the field of view of the large meri- A!manac. ‘I’he observations were among the first under- I dian photometer is about forty minutes of arc, and the taken and were made with the sm:tll meridian ,photometer. position of the star to be observed could ordinarily be set In general, only hright stars were observed with this in- I upon to within ahout five minutes of arc. Observations strument, since it WAS feared that errors in identification I of ten equatorial stars, selected at random, g.ive an average might occur when the star was faint. Several such cases I deviation in their recorded positions of two minutes of arc are pointed out in Volume XIV, and, like this one, gene- I in , and about four minutes of arc or sixteen rally arose from errors in other catalogues. I seconds of time in right ascension. The star should thcre- Though this is the only instance given as regards I fore appear near the centre of the field of view, and ))right Volume XIV, we are assured that the discordances are stars generally required no further identification. If. how- \)startling and numerous and have puuled astronomers( ! ever, the observer was in doubt, or thought that on another and are thcw told that xto go into the errors of the Photo- occasion a inistake was likely to be made, he always gave I metric Catnlogue lies outside of the intended scope of this the position of one or more adjacent stars. On the foll- notec. We will only say here that a cornparison (€€;ward I owing day these stars were compared with the catalogues, Annals, Vol. XVIII, p 13) of the measures of a large number and a careful examination made to determine whether the of stars with those obtained at three other observatories, I star observed was the one intended. Many thousands of as stated in niore detail below, shows that in one case I stars were thus recorded, both with the small and the only, H. 1’. 2.179, the nieasures made here differ from the large meridian photoineters, and the amount of time spent I mean of 311 by more than three tenths of a magnitude. on their identification was very great. With the larger In [his c,m, our reciults are confirnied by the estimates of i instrument care was also taken to record the exact position W. Herschel and Schonfeld. and also by other measures I of the mirror reflecting the star into the field of view, as made at this observatory with a different instrument. Doubt- I well as the time of observation, so that the approximate less there may be other errors in Volume XIV like that I right ascension of the object observed could be determined. mentioned above, which repeated observations and a wider I ‘l*his additional deterrnination of the position of the star 15 219 3229 2 20

was made the next day in all doubtful cases. 'I'he setting ' was much less. When a was so faint as to in declination, as shown above, could be made with such 1 be invisible in the photometer, it sometimes happened that accuracy that before 1889 it was not thought necessary to i an adjiicent faint star was observed by mistake. Such record the reading of the declination circle in each case. I errors in identification are liable to occur with any instru- Since that time, these readings have been made, so that 1 ment. In such cases, a large difference in brightness is the right ascension and declination of the object observed as probable as a small one, and would never be used to can always be determined from the record. 1 indicate the amount of error in measurement hy a critic When faint stars were to be measured, additional I who was imprejudiced. It is as if a rifleman should aim precautions were taken. They were generally observed in 1 at a wrong target, and the distance between the two targets zones having a width of twenty minutes, and the position 1 be taken as a measure of his marksmanship. of each was verified by that of the preceding and following I It h3s always been the policy of the observatory to star. As an example, after measuring one star the observer print all the otmrrations made with the meridian photo- would be informed by the recorder that the next star foll- ' meter, in order that the reader might have in his possession awed one minute and five tenths and was ten minutes I the entire material. A full discussion of those which appear north. He then estimated its rriagnitttde, and the catalogue I to be in error is afterwards made. Unfortunately, the magnitude was read to him I)y the recorder. A large I volume in which the observations of variable stars arc to difference would suggest error and one or more adjacent ' be discussed has not yet been published. stars would be inserted for subsequent identification. The ' Prom a re-examination of the original records, it recorder would then tell him that the next star followed I appears that for R Andromedae, the adjacent star BD. six tenths of a minute and was fifteen minutes south. The +37?57 was observed; for Y Virginis, BL). -33"3313; for obserrer thus doubly checked the position of each star, ' 2 Virginis, BD. - I 203980 (not BD. - 1203993 which has leaving little chance for an error in identification. In this I the magnitude 8.8 not 8.0, as given by Mr. Chandler) ; for way three independent checks were applied to insure the I S Scorpii, BL). -2204144; for ... Serpentarii, BD. - z 104626; correctness of the identification of the stars observed ; I for R I.acertae, RD. +4104503 (not 131). +41?4j91) ; for first, the settings in right ascension and declination ; second, j S Aquarii, BE). - 2 106334 ; for R Pegasi HD. c905 I 59. In the configiiration of adjacent stars; and third, the differ- these eight cases then the error was not in the plan of ences in right ascension and declination of the star oh- ' observation, which actually provided means for the satis- served from those of the preceding and following stars. I factory identification of the star observed, but occurred io

The record thus shows with certainty what star wa,s mea- ~ the original reduction. The observations of 'r Leonis, and sured in almost every case, except in the earliest obser- i of X Virginis were among the first made with the large vations. 'I'hese observations, unlike most others of a meridian photometer, xnd are probably in error. 'The ori- similar kind, do not therefore leave us rat the mercy of ' ginal record of the observations of S Pisciutn in 1888 is the record< as Mr. Chandler supposes. In reality, no portion I missing. The star observed was seen with difficulty on of the work was more laborious than the identification, I account of moonlight. Photographs show a star so near both in observation and in reduction. , the place of 'r Coronae that the magnifying power used The varhble stars referred to on page 355 of the I with the meridian 1)hotometer would not make the difference Nachrichten, and used as a criterion of the value of the i in position perceptible. A similar remark aplllies to the whole work, were among the most difficult objects whose I case of 'r Scorpii; the centre of the cluster NGC. 6093 observation was attempted. They were generally very faint, , was observed. The original record of the observation of or entirely invisible in the photometer, and the best evi- I ... Ophiuchi on July 25, 1886 was >not seen, fainter than dence that they were correctly observed is their relation 8.8~; on the only other date of observation El). - 1204634 to Adjacent stars. So much light is lost by the polarizing was the object observed. 3ppratus that closely adjacent faint stars cannot be seen, It cannot be assumed that observations of variable and brighter and more distant companions must be used. I stars of long period like V Coronae are wrong because 'I'he difficulties to be overcome in such cases are as great , their measured magnitudes do not agree with those found as those encountered by observers with ordinary telescopes I by prediction. The best known object of this class is on moonlight nights. As these variables are scattered over i o Ceti for which we have observations extending over more all parts of the sky, they could not, like other faint stars, I than four centuries. Yet its time of mAximum this , be observed in zones, and identified with regard to one as predicted by Mr. Chmdler, was to be February 17, a I another. That care in identification was taken, is shown ' date more than a month in error since the maximum by the fact that in preparing the list adjacent stars were actually occurred in the latter part of March. The bright- entered in each case on the observing shects from the I ness of this star differs more than three magnitudes at Catalogiie. In addition, others were in- ! different maxima, and since similar differences occur in serted at the time of observation. 'I'he latter generally I other variables it is not surprising that the observations of enable us in doubtful cases to decide with certainty which V Coronae, with the meridian photometer, do not agree star wcis observed. i with prediction, but show that the star was a magnitude It is obviously unjust, as we have above urged, to fainter at maximum in 1888 than in 1887. 'I'his star was use .the observations of such objects as a test of the accu- ! observed on eight nights in 1887, and on nine nights in racy uf other obseriations in which the danger of error . 1888, and its position was checked by two adjacent stars. 22 I 3229 222 lhese observations give a total change in brightness of a I than two tenths of a magnitude are 20, 15, 26, and 6 magnitude and a half. They are all represented with an , respectively, and that the average deviations of each from average residual of 0.14, by assuming that the increase of I the mean are 0.10, 0.09, 0.10, and 0.06 respectively light was uniform during each year. (Vol. XVIII, p. 13). The conclusion that the average error The vagueness of the charges made by Mr. Chandler, j of stars in the Harvard Photoinetry does not exceed one that numerous errors occur in the r)hotomctric cataloeuesu tenth of a magnitude, therefore seems justified (Vol. XIV, contained in Volurnes XIV, and XXIV, makes it .difficult 1). 505). A similar coniparison of 58 circumpolar stars, to refute them. It is impossible to prepare an extensive measured by Professor Ceraski of Moscow, gave an average catalogue without error. Many errors have been found in difference of 0.14 of a magnitude, and shows that the all catalogues hitherto published The general accuracy of .average error of the magnitudes given in Volume XXIV, the work, both in measureiiient and identification, is best also is not likely to exceed one tenth of ;I magnitude. indicated by comparing the results of different nights with This result is confirrnpd by comparison with various other one another, and with those obtained at other observatories. observations including those made at the Pulkowa Obser-

'I'he smallest difference between two stars that is percep- I vatory by Dr. E. Lindemann. tible to the eye, is between one and two tenths of a ! The accordance of the results ohtained in different magnitude. 'I'he average deviation of a single ohservation I with the large meridian photometer is shown in in the Harvard Photometry amounts to about 0.16 of a I Volume XXIII, p. 64. 'I'he average deviation of the annual magnitude (Vol. XIV, 1). 35). In the second instrument used, ' results of one hiindred stars observed from 1882 to 1888 with which the observations now under discussion were , is 0.04 of a magnitude. l'he average value of the resi- rnitde, this difference was reduced to ahout 0.12 (Volume I duals included in Volume XXIV, but slightly exceeds one XXIll, p. 44). 'The computed probable error in the deter- I tenth of 3 magnitude. inination of the brightness of :I single star in the Harvard It is therefore claimed that the meridian photonieter

Photometry is 0.072 of a magnitude (Vol. XIV. 1). 497). ~ is both an accurate and a rapid means of measuring the The results of this work, as cornpared with the most accu- brightness of the stars. During the last fifteen years cata-

rate catalogues previously published, show a difference on ~ logues of 4260 stars have been published in Voluine XIV, thr average of less than two tenths of a magnitude. This and of 20982 stars in Volumes XXIII and XXIV. The includes the errors of both the catalogiies, in addition to 1 ringnitudes of 7922 southern stars arc in print in Volume the ainorint of their systematic differences (Vol. XIV, 1) 502). XXXIV, but are not yet published, and about seven thousand A later conip:irison with the photometric: ca:alogurs of stars have been measured since 1891. Professor Pritchard of Oxford, and of Dr. Wolff of Bonn, , 'These include the re - observation, now nearly com- indicates that the results of the Harvard Photoinetry dtffers I pleted, of the stars in the Harvard Phimrnctry, with the

from these two catalogues by 0.140and 0.146 magnitudes ' large meridian photometer. Any remaining errors in that

rcspectively (Vol. XVILI, 1). 27:. 'l'he inngnitudes of the ~ work can thus scarcely fail to be detected. About 560000 Uranonietria Argentina, those obtained by Wolff and by ' photometric settings have, so far, been made with the Pritchard which arc mentioned above, and those in the ' meridian photometers. The most extensive photometric Harvard Photometry are cornpsired in Volume XVILI, No. I. I catalogue published elsewhere is that of Professor Pritchard, After reducing them to the same scale it appears that the I which contains 2784 stars. numbers of the stars that differ from the mean hy inore

Harvard College Observatory, Cambridge, Mass., I 894 .4pril I 8. Edward C. Pickering.

Uetw den Einfluvs der Spa1t)weite auf das Aiisselien der Cornetenspectra. Von H. KaysPr.

In Nr. 3221 der A. N. hat Herr Pro6 H. C. Vogel Be- I gezogert haben, rneine Betrachtungen zu veriiffentlichen, I merkungen an meinen hrtikel in den A. N. 32 17 unter denn dieselben cnthalten wesentlich metir, als obiger Passus obiger Ueherschrift geknupft, auf welche ich P'olgendes zu I von Vogel. Er verbffentlicht nur eine Beobachtung, dass

erwidern habe. I bei breitem Spalt auch beim gewohnlichen Kohlenspectrum Ich bedaure lebhaft, dass mir der von Vogel zum I das Lichtinaximum von der Kante fortrucke; aber eine Wiederabdruck gebrachte l'assus einer alteren Abhandlung I Erklarung und weitere Erforschung des Einflusses der Spalt- (Publ. d. Astrophys. Obs. Potsdam 11, 1881) in der That bei I weite giebt er nicht. Bei mir dagegen findet sich grade meinem Stuclium der Litteratur uber Cometenspectra ent- , diese Erklarung, aus welcher denn auch sofort eine Reihe gangen ist. Ich wurde sonst nicht verfehlt haben, ihn zu weiterer Erschcinungen, welche Vogel nicht beobachtet hat, citiren und hervorznheben, dass Vogel der crste gewesen, I und die Bedingungen fur ihr Auftreten sich ergeben. Ich der die Bedeutung der Spaltweite erkannt hat. I muss daher den Ausspruch Vogel's, rnein Aufsatz enthalte

Andercrseits wlirde icii trotzdem keinen Augenblick I nur die Bestatigung von Thatsachen, die dem Kenner der 15: