May 31, 1894] 



NA TURE 



lO- 



the geographical latitude. If, on the other hand, the trails of 

 the two stars do not touch, their shortest distance may be 

 measured. Now the distance of the trails of any other two stars, 

 which appear during the same exposure, will, when their declina- 

 tions are taken from the almanacks, enable us to calculate what 

 angle approximately corresponds to a certain length, the 

 approximation being close for small lengths. Thus we are 

 enabled to calculate what the declination of the second star 

 would have been, if its trail had touched the trail of the first 

 star, and thus the latitude may be found. 



Generally there will be the trails of several stars in both ex- 

 posures available, so that one plate will allow of several deter- 

 minations, and thereby of an elimination of errors. The results 

 of seven plates, taken on different nights in my garden in Han- 

 nover, Germany, will give an idea of the accuracy. 



52^ 22' -50 

 52° 22''S6 

 52" 22'-47 



52" 22'78 



52" 22' 90 



52° 22'-93 

 52° 22''90 



Mean: 52' 22'72 (mean error of a single 

 plate =u''2). 



According to the Pretissische Landesaufnahme the latitude 

 should be — 



52° 22''84 



The same plate may also serve to determine the deviation of a 

 watch from local mean time, if the time of the watch is noted 

 each time the cap is withdrawn and replaced. Imagine, first, 

 that a star was photographed whose declination coincided with 

 the geographical latitude, and imagine that the box could be 

 turned infinitely quickly without any oscillations being started, 

 ff we then found that the two trails of the star corresponding to 

 the two exposures touched each other at one end, the end corre- 

 sponding to the moment the box was turned, we might conclude 

 that the star at this moment was in the zenith. For there alone 

 the image of the star would remain unaffected by the turning of 

 the plate. Now, we can calculate from the almanack the 

 local mean time of the moment when this particular star cul- 

 minates,- and this would give us the deviation of our watch 

 from local mean time. Imagine now the box had not 

 been turned at the moment the star culminated, but, say, 

 ten minutes later. Then the two trails would not have one end 

 in common. The ends corresponding to the moment when 

 the box was turned, would be on opposite sides of the zenith 

 at the same distance from it. Now we should be able by the 

 length of the trails to determine what interval of time corre- 

 sponds to a given length, and so the plate would show us that 

 the culmination of the star took place ten minutes before the 

 box was turned. If the box is not turned very quickly, the in- 

 terval after replacing and before removing the cap must be 

 taken into account. But it is evident that we can calculate 

 where the ends of the trails would have been if the box had 

 been turned infinitely quickly. If there is no star that passes 

 exactly through the zenith, we must take one that passes near 

 the zenith. This will also do. By the ends of the trails we 

 find which point of the plate corresponds to the zenith, and at 

 the same time we can, by the trails and their ends, find out the 

 line representing the meridian. This gives us the necessary 

 data for finding the deviation of our watch from local mean 

 time. For a star near to the zenith a small error in the direc- 

 tion of the meridian will make very little difference. I have 

 spoken of the ends of the trails as a means of determining local 

 mean time only for the sake of simplicity. It is far preferable 

 to proceed differently, and to interrupt the first and the second ex- 

 posure a number of times, thus causing as many interruptions of 

 each trail. Five seconds or less will do very well for the length 

 of an interruption. These interruptions are preferable to the 

 ends, because they are symmetrical, and their middle can with 

 more accuracy be brought under the crosshairs of a micro- 

 meter. My lens having a focus of about 24 cm. made one 

 minute about equal to GO mm. on the trails of stars that cul- 

 minate near the zenith. With a micrometer it is not diflicult 

 to measure ex.ict within 001 mm. which is ei|uivalent to one 

 second of time. 



If by the help of a chronometer Greenwich mean time were 

 known at the place where the photograph was taken, the 



NO. I 283, VOL. 50] 



longitude would thus also be determined. But a chronometer 

 can be dispensed with if another photograph is taken of the 

 moon and the stars. For this purpose the camera is taken out 

 of its bo.x, directed to the moon, and fixed as firmly as possible. 

 A number of instantaneous photographs of the lunar disc are 

 then taken, all on the same plate, at intervals of not less than 

 two minutes, to prevent the images from overlapping. The 

 camera must be touched as little and as gently as possible, so 

 that it may remain in the same position. After a number of 

 exposures, say six or eight, the camera is shut unlil the moon 

 has quite gone out of the field. The cap is then again 

 removed, and the stars draw their trails over the plate, interrupted 

 here and there, the lens being now and then covered for five 

 seconds. If the local mean lime of the instantaneous exposures 

 of the lunar disc and of all the interruptions is known, the 

 Greenwich mean lime may be found from the plate. One may 

 apply three different methods. Either one can determine the right 

 ascension of the moon or the declination, or one can measure 

 lunar distances. I have tried all three, and prefer the second, 

 provided the moon is not near the maximum or minimum of her 

 declination. For the slower rale of change of declination 

 is made up by the greater accuracy with which it can be 

 measured, on account of the trails being lines of constant declina- 

 tion. Each image of the lunar disc allows a separate deter- 

 mination, so that the final value is the mean of a number of 

 determinations. (For the details of measuring, see my article in 

 Xhe Zeitschrifl fiir Vermessungiwesen, August 1S93.) 



The results of one plate for the dift'erence of Greenwich mean 

 time and local mean time were by the three methods — 



39'I minutes 



Mean 38-93 „ 



while the true value is 38'943. This very close coincidence 

 must be considered accidental. But t think one can well rely 

 on the determination from one plate not differing more than 0'2 

 minute from the real value. When exposing for a longer time, 

 the dew will sometimes condense on the lens, and render it 

 opaque. To prevent this, I place a screen before the lens with 

 a hole in it a little larger than the lens. The screen keeps 

 the lens from cooling below the temperature of the surround- 

 ing air, and thus removes the cause for the condensation of dew. 

 When determining the latiiude the screen was arranged as a lid 

 of the outer box, and served at the same time lo protect the 

 camera from gusts of wind. C. RUNGE. 



Technische Hochschule, Hannover. 



Sodium and Uranium Peroxides, &c. 



In your notes of May 17 you give an account of work done 

 by Prof. Poleck on the action of sodium dioxide on the salts 

 of various meials, from the BerUhte of May 8. 



Some of this work has been already recorded by other 

 observers who used either sodium dioxide or hydrogen dioxide 

 in alkaline solutions. 



In the yw/rHd/ of the Chemical Society, 1S77, pp. 1-24, and 

 pp. 125-143, I gave papers on the reactions of hydrogen dio- 

 xide, and described its action on uranium salts both in acid and 

 in alkaline solutions. I do not wish, however, to say anything 

 on a mere question ol priority, but simply to point out in- 

 accuracies in the formulae given in Nature from the Biiichtc. 



Prof. Poleck gives for sodium peruranate the formula 

 Na^U.O, . 8H.,0. This either involves the use of the old 

 atomic weight of uranium, half of that now adopted, or it means 

 that his salt is exactly like that prepared by me in every respect, 

 excepting that it contains twice the proportion of uranium. 

 Assuming the atomic weight at approximately 240, the correct 

 formula of sodium peruranate is NajUO, . SH.jO (Journal of 

 the Chemical Society, 1877, p. 139). 



I also showed that hydratcd sodium dioxide is readily 

 obtained on adding alcohol to mixed sodium hydrate and 

 hydrogen dioxide solutions ^p. 125), so that the materials used 

 by Prof Poleck and myself are practically the same. 



The uranium compounds ate interesting as examples of a 

 special highly oxidised type (see Mendelueff's " Principles of 

 Chemistry," vol. ii. p. 244), also as throwing light on the nature 

 of certain analogous but unstable bodies, such as perchromic acid. 

 (yoKrwd/ of the Chemical Society, 1877, pp. 7-8I. This was 



