The computations give the following table* and curve which show the variation of latitude.

The curve required by Dr. S. C. Chandler's formula (Ast. Jour., No. 446) is shown in the dotted line. From 1896 the observed epochs of maxima and minima follow the computed in time.

Dr. Chandler writes me that "on account of the different combinations of the observers participating at various times, there are doubtless constant differences of zeros in the observed curve at different epochs, which would bodily shift the latter up or down, over certain intervals. However the noticeable systematic differences of the observed and computed curves (notably from 1896 to 1898 where the observed epochs of maxima and minima distinctly follow the computed in time) are undoubtedly real, as well as extremely interesting. In my opinion they are referable to the fact that the annual ellipse is not stationary, but that its line of apsides is retrograding several degrees annually, as I have pointed out in my articles in various places (see especially Ast. Journal No. 446 where I have summed up the evidence). Of course the law of this revolution is not yet sufficiently demonstrated to justify its being incorporated in the formula which is hence based on a constant position of the ellipse (39°.6); that is, the inclination of the major axis of the annual ellipse is assumed as 40° east of Greenwich, although the evidence is sure that it is backing around, so that it is now (1898-99) somewhat west of Greenwich. The fuller and more satisfactory investigation of this phenomenon must await the accumulation of two or three years' observations. By 1902 or 1903 I think it can be clearly apprehended, and formulated; not before. Another interesting fact is that your later observations confirm what your and other series earlier indicated, namely, that the dimension of the fourteen months' circular motion has progressively diminished since 1890. This is so well established that it was embodied in the formula some years ago." Nearly every element entering into this motion of the pole is variable. The four series of observation gave the foregoing values of the Aberration Constant:

Taking the probable error of a single latitude observation as

* The previous publications will be found in The Astronomical Journal Nos. 401, 451, 474. In these papers the mean latitude is taken for each series and not, as in this case, for the whole time.

0'.16 gives Constant of Aberration 20".464 ± 0.006.

Fergola obtained a value, corresponding in time to our Series A, of 20".53, using the same stars and the same methods of reduction.

Harkness in his work on "The Solar Parallax and its related Constants" (1891) gives the values of the constant of aberration determined between the years 1817-1888, and from his discussion of these values gets 20".466 ± 0.011. Newcomb in 1895 published his book on "Astronomical Constants" in which he gave the values of the aberration constant down to 1894. He divided the results into two classes: A, standard Pulkova determinations, and B, other determinations. The separate results he weighted and obtained

The first result is almost "identical with that found by Nyrén" in 1883: this was 20".492 ± 0′′.006 and was not included in list A or B.

These results were all obtained on the assumption of a constant latitude. Dr. Chandler has rediscussed and obtained quite different results. The Paris Conference of May 1896 adopted the value of 20".47.

Doolittle's results gave him a mean value of 20".580 from observations made 1896-1898: Fergola obtained, as stated above, at Naples the value 20′′.533. Finlay at the Cape of Good Hope obtained the value 20".57 and at Berlin and Strassburg have been obtained the values 20".511 and 20".475.

The value 20".47 corresponds to the value of solar parallax of 8".8033, while 20".53 gives 8".7773.

The differences in the various results obtained indicate plainly that we must wait some time longer for a definitive value of the aberration constant.

PHOTOGRAPHS OF THE ZODIACAL LIGHT.

A. E. DOUGLASS.

FOR POPULAR ASTRONOMY.

The experiments which resulted in the accompanying photographs of the zodiacal light were the outcome of long continued interest in the subject of the gegenschein and zodiacal light and a desire to render less fatigueing and more accurate the observation of those faint lights. Contours of the zodiacal light drawn

by hand on about two hundred different nights had shown the defects and difficulties of that method of recording observations. Perhaps the chief of the defects was the almost invariable interference of stellar light; yet that was not the only important one, for the very first result of these photographs was to show that in estimating a visual contour one is apt to mistake ease in distinguishing the zodiacal cone at any given point for actual intensity at that point and therefore the contours of equal brightness are prolonged along the axis too far away from the Sun and horizon light. After seeing the photographs this erroneous tendency was actually found to exist.

The inherent difficulties in the old method lie in the long time required to get perfectly acquainted with standard reference stars. and the advantage, that amounts almost to necessity, in making records in the dark. Even an old hand at this work will frequently be obliged to make use of some star whose name he does not know and will have to describe its place, as well as that of the zodiacal light, on paper which he can barely see resting in his hand.

The first effort to improve upon the method of observation consisted in designing a machine which could automatically record the contours directly on a star map, and a rough model of this contrivance gives entire promise of success. But as this was not put into actual practice attempts were made to succeed by photography. Many attempts have been in this line elsewhere, but without success. Of these one was by myself while in South America (in 1891, I think) and another has since then been made at that same station. The early work at Flagstaff was equally without result. But after repeated trials success was at once attained when the very simple idea of using ordinary positive visual eyepieces, which combine very short focus with relatively large aperture, occurred to me and was tested and from that time on constantly improving results have been obtained. The most successful lens and the one by which the illustrations for this article were taken, is a combination lens, put together and mounted by Mr. Cogshall, who has done practically all of the photographic work. This lens seems to give a combination of flat and large field and very great light-transmission that surpasses any other apparatus that we have tried, and among those tried the one marked "Clark; Special" in the list below is a "solid" achromatic lens cut on the Fraunhofer curves. The following list describes briefly the various lenses tried by Mr. Cogshall and gives a summary of our experiments:

After obtaining real photographs the first step was to make sure that there was no deception. Some of the views taken as long ago as May, 1899, were really exceedingly good but had some trifling defect, which threw a slight doubt on their genuineness. But they were repeated many times. Finally a very thorough test for ghosts or other concentration of light in the fields of the chief lenses used, was made by trial exposures on such objects as landscapes and ruled paper, by daylight, the side of a house by moonlight, and the sky, by day and by night. No genuine irregularity or concentration of light could be found in the fields over an area considerably greater than the entire portion of sky shown in our illustrations, and the photographs were therefore accepted as real. (Let me here explain that while taking the photograph of October 7, a pasteboard tube supporting the lens, projected too far inside the camera and cut down the field; this effect shows conspicuously in the original negative.)

The first conclusions drawn from these photographs are that the axis of greatest density is very indefinite and that the photographic contours of the apex of the zodiacal cone are far more rounded in form than the visual ones are usually represented; and, as stated above, I am inclined to think this same roundness is really true of the visual outlines.

Another conclusion and one of much interest to the experimenter, is one that cannot be derived from our illustrations but has been found to hold true on other photographs. It is that with this form of camera the zodiacal light makes an impression on the sensitive plate more readily than equally bright regiors of the Milky Way.

The horizon light does not appear to effect these photographs.