SeI'TEMDER 



■/ ' 



1S94] 



NA TURE 



03/ 



all Ihese correclions must be so determined as to make ihe sum 

 of their weighted squares a minimum, and at the same time 

 satisfy every one ol the equations expressing the relations 

 between the various components of the group. 



Thus it appears that the method required for adjusting the 

 solar ]iarallax and its related constants is in all respects the 

 same as that which has so long been used for adjusting systems 

 of iriangulation ; and as the latter method «as invented liy 

 astronomers, it is natural to inquire why they have not applied 

 it to the fundamental problem of their own science? The 

 reasons are various, but they may all be classed under two 

 heads. First, an inveterate habit of over-estimating the 

 accuracy of our own woik as compared with that of others ; 

 and second, the unfortunate effect of too much specialisation. 



The prevailing opinion certainly is that great advances have 

 recently been made in astronomy, and so they have in the fields 

 of spectrum analysis and in the measurements of minute quanti- 

 ties of radiant heat ; but the solution of the vast majority of 

 astronomical problems depends upon the exact measurement of 

 angles, and in that little or no progiess has been made. 

 Bradley, with his zenith sector, a hundred and fifty years ago, 

 and Bessel and Struve, with their circles and transit instruments, 

 seventy ) ears ago, made observations not sensibly inferior to 

 those of the present day, and indeed it would have been sur- 

 prising if they had not done so. The essentials for accurately 

 determining star places aie a skilled observer, a ciock, and a 

 transit circle, the latter consisting of a telescope, a divided 

 circle, and four micrometer microscopes. Surely no one will 

 c'aim that we have to-day any more skilful observers than were 

 Bessel, Bradley, and Struve, and the only way in which we have 

 improved upon the telescopes made hy iJollond one hundred 

 rxnd thiity years ago, is by increasing their aperture and re- 

 latively diminishing their focal distance. The most famous 

 dividing engine now in existence was made by the elder Kepsold 

 seventy-five years ago ; but as the errors of divided circles and 

 their micrometer microscopes are ahvays carefully determined, 

 the accuracy of the measured angles is quite independent of any 

 small improvement in the accuracy of the division or of tlie 

 micrometer screws. Only in the matter of clocks has there 

 been some advance, and even that is not very great. On the 

 whole, the star places of to-day are a little better than those of 

 seventy-five years ago, but even yet there is great room for 

 improvement. (Jne of the commonest applications of these 

 star places is to the determination of latitude, but it is very 

 doubtful if there is any point on the face of the earth whose 

 latitude is known certainly within one-tenth of a second. 



Looking at the question from another point of view, it is 

 notorious that the contact observations of the transits of Venus 

 in 1761 and 1769 were so discordant that from the same obser- 

 vations Kncke and E. J. Stone got respectively for the solar 

 parallax S 59 seconds and 8 91 seconds. In 1.S70 no one 

 thought it possible that there could be any such difhculty with 

 the contact observations of then approaching transits of 1S74 

 and 1SS2, but now we have found from sad experience that our 

 vaunted modern instruments gave very little better results for 

 the last pair of transits than our predecessors obtained with 

 much cruder appliances in 1761 and 1769. 



The theory of probability and uniform experience alike show 

 that the limit of accuracy attainable with any instrument is soon 

 reached ; and yet we all know the fascination which continu- 

 ally lures us on in our efforts to get better results out of the 

 familiar telescopes and circles which have constituted the 

 standard equipment of observatories for nearly a century. Pos- 

 sibly these instruments may be capable of indicating somewhat 

 smaller quantities than we have hitherto succeeded in measuring 

 with them ; but their limit cannot -be far off, because they 

 already show the disturbing effects of slight inequalities of tem- 

 perature and other uncontrollable causes. So far as these effects 

 ^re accidental, they eliminate themselves from every long series 

 of observations, but there always remains a residuum of constant 

 error, perhaps quite unsuspected, which gives us no end of 

 trouble. Encke's value of the solar parallax affords a fine 

 illustration of this. From the transits of Venus in I 761 and 

 1769 he found S5S seconds in 1S24, which he subsequently 

 corrected to S'57 seconds, and for thirty years that value was 

 universally accepted. The first objection to it came from 

 Hansen in 1S54, a second followed from Le Verrier in 1S5S, 

 both based upon facts connected with the lunar theory, and 

 eventually it became evident that Encke's parallax was about 

 one quarter of a second too small. Now please observe that 



NO. T3OO, VOL. 50] 



Encke's value was obtained trigonometrically, and its inaccuracy 

 was never suspected until it was revealed by gravitational 

 methods which were themselves in error about one-tenth of a 

 second, and required subsequent correction in other ways. 

 Here then was a lesson to astronomers, who are all more or 

 less specialists, but it merely enforced the perfectly well-known 

 principle that the constant errors of any one method are ac- 

 cidental errors with respect to all other methods, and therefore 

 the readiest w ay of eliminating them is by combining the results 

 from as many different methods as possible. However, the 

 abler the specialist the more certain he is to be blind 10 all 

 methods but his own, and astronomers have profited so little 

 by the Encke-Hansen-Le Verrier incident of thirty-five years 

 ago that to-day they are mostly divided into two great parties, 

 one of whom holds that the parallax can be best determined 

 from a combination of the constant of aberration with the 

 velocity of light, and the other believes only in the results of 

 heliometer measurements upon asteroids. By all means con- 

 tinue the heliometer measurements, and do everything possible 

 to clear up the mystery which now surrounds the constant of 

 aberration ; but why ignore the work of predecessors who were 

 quite as able as ourselves? If it were desired 10 determine 

 some one angle of a triangulation net with special exactness, 

 what would lie thought of a man who attempted to do so by 

 repeated measurements of the angle in question, while he per- 

 sistently neglected to adjust the net ? And yet, until recently, 

 astronomers have been doicg precisely that kind of thing with 

 the solar parallax. I do not think there is any exaggeration ir. 

 saying that the trustworthy ob.ervations now on record for the 

 determination of the numerous quantities which are functions 

 of the parallax could not be duplicated by the most industrious 

 astronomer working continuously for a thousand years. How- 

 then can we suppose that the result properly deducible from 

 them can be materially aflected by anything that any of us can 

 do in a lifetime, unless we are fortunate enough to invent 

 methods of measurtment vastly superior to any hitherto 

 imagined? Probably the existing observations for the deter- 

 mination of most of these quantities are as exact as any that 

 can ever be made with our present instruments, and if they 

 were freed from constant errors they would ceitainly give results 

 very near the truth. To that end we have only to form a 

 system of simultaneous equations between all the observed 

 quantities, and then deduce the most probable values of these 

 quantities by the method of least squares. Perhaps some of 

 you may think that the value so obtained for the solar parallax 

 would depend largely upon the relative weights assigned to the 

 various quantities, but such is not the case. With almost any 

 possible system ol weights the solar parallax will come out very 

 nearly S'Sog seconds ± OC057 seconds, whence we have for 

 the mean distance between the earth and sun 92,797,000 miles, 

 with a probable eiror of only 59,7CO miles ; and for the diameter 

 of the solar system, measured to its outermost member, the 

 Planet Neptune, 5,578,400,000 miles. 



THE METEOR AND METEOR-STREAK OF 

 AUGUST 26, 1S94. 



■T^HE present year will certainly be remarkable for its large 

 -'- meteors. One of the most brilliant class of these phe- 

 nomena appeared on January 25, and a fortnight later (February 

 8) a fireball was seen as a conspicuous object even in the 

 pi esence of the midday sun, for the time was only 28 minutes 

 after noon. The early evening of February 21 furnished another 

 of these brilliant objects, but the observations were neither 

 numerous nor exact, and all that could be definitely gleaned 

 from them was that the body disappeared at a height of 30 miles 

 over Bolton in Lancashire. On April 22, before daylight had 

 gone, a fine meteor descended over the extreme south-east part 

 of England, crossing Ihe Strait of Dover from Hastings in the 

 direction of Amiens in France. On May iS a large daylight 

 meteor was observed in .Scotland and Ireland. Several addi- 

 tional instances of these striking visitors have been recently re- 

 corded, and the Perseids presented a few fine specimens, thoug*- 

 the season has been a very cloudy and unpropitious cne for all 

 kinds ol celestial observation. 



The magnificent meteor which forms the subject of this paper, 

 appeared on August 26 at loh. 2oni. It did not owe its parent- 

 age to the great Perseid system, for it came too late in the 

 month, and, moreover, its direction of flight is not conform.ible. 



