Sept. 5, 1889] 



NATURE 



44: 



clocks), may be divided into two classes— (a) those in 

 which uniform motion is obtained, or sought to be 

 obtained, by some variable friction or resistance which 

 increases as the speed increases ; and {b) those in which 

 some such similar contrivance is supplemented by a 

 system of electric control from a pendulum clock, which 

 is itself incapable of being re-acted upon by the uniform 

 motion clock. 



For all ordinary observing, and even for micrometric 

 •work, clocks of the class a are made, which answer 

 admirably, but for photographic equatorials I believe it 

 will be found necessary to employ clocks of the b type, 

 and for this reason : the tendency of the compensation in 

 uncontrolled clocks (class a) is to correct the rate of the 

 clock when from any momentary cause it is disturbed. 

 The best it can do is to bring the rate absolutely right 

 again, but it cannot act till an error has actually occurred, 

 and therefore, although the rate is corrected, the posititm 

 of the star on the plate is shifted by the amount of the 

 error. I have heard it stated by the designers of some 

 of these clocks that errors were corrected before they 

 existed ! It is hardly necessary to stop to show the 

 fallacy of this, but it is evident that the increased or 

 .diminished resistance, or friction, or whatever it is, that 

 •checks the speed, can only exist from, and in consequence 

 of, the error itself. 



In the case of micrometric measures, it is not of very 

 niiuch consequence if a minute error occasionally creeps 

 in, provided the speed keeps constant during the few 

 seconds or minutes required for making the bisections, 

 but in the case of the photographic telescopes, if the 

 image of the star takes up a new position anytime during 

 the exposure, it is of course fatal. 



Let us try now and get an idea of what amount of 

 accuracy is really necessary for this work. We often 

 hear .of a perfect equatorial clock, but the word perfect is, 

 I fear, as loosely used in this connection as in others. 



I have heard in days gone by a perfect clock defined 

 as one which drove the instrument so accurately that if 

 you set the telescope on a star and went to dinner you 

 would find the star still in the field when you resumed 

 your observations after dinner. Allowing, say, two hours 

 between the ante -prandial and the post-prandial observa- 

 tions, and assuming the eye-piece to be (as we may 

 fairly do) a low one of about 20' of arc field, this 

 would mean that the clock did not vary more than 600" 

 of arc or 40 seconds of time an hour. The accuracy we 

 now require for these photographic telescopes is some- 

 thing very different. The image of a 12th magnitude 

 star impresses itself on the plate, with moderate exposure, 

 in the form of a circular disk of about f\^ inch diameter. 

 If the clock vary one-tenth of a second during the 

 exposure, the disk will be elongated by i^Vir inch, 

 producing a very sensible distortion. 



We must not therefore have any errors over one-tenth 

 of a second, and if possible it should be reduced to one- 

 twentieth. 



It will not be necessary that the clock keep within this 

 one-twentieth of a second for more than ten or fifteen 

 minutes, because it is always necessary to watch the 

 image occasionally through the guiding telescope, and 

 correct whenever refraction becomes apparent ; but what 

 I do urge as absolutely necessary is that the clock shall 

 go so perfectly as not to require more than the occasional 

 attention of the observer, instead of the constant and 

 never-ceasing watching with ordinary clockwork. No 

 one who has not tried it can imagine the strain required 

 to keep a constant watch on a star image for 30 or 40 

 minutes, but if attention be only required for a second or 

 so every few minutes there is no difficulty or irksomeness 

 whatever. 



Even the most enthusiastic admirers of various forms 

 of equatorial clocks will not venture to assert that they 

 will go for fifteen minutes without one-tenth of a second 



of error. There is now, however, no difficulty in con- 

 trolling a uniform motion clock from a pendulum so 

 that it will never vary one twentieth of a second from it. 

 it may therefore, I think, be assumed that some form of 

 electrical control is necessary. There are, as far as I 

 know, four forms of control to choose from. 



First, Dr. Gill's, as applied to the 15-inch equatorial at 

 Dun Echt with admirable success. 



In this an electric current is sent once a second from 

 an independent pendulum, which may be any distance 

 away. That current passes through a certain wheel in 

 the clock, with contacts so arranged that if the clock be 

 going exactly with the pendulum the current is sent in*a 

 direction which keeps one of two rubbers rubbing on a 



Fir.. I. 



quick-moving wheel of the clock. If the clock, however, 

 goes the least quantity too fast, the wheel has revolved 

 a little further than it should at the moment the next 

 current comes from the pendulum, and the current is sent 

 in such a direction as to cause both rubbers to rub on 

 the clock wheel. If, on the contrary, the clock has gone 

 a shade slower, the current is sent in a third direction, 

 which lifts both rubbers oft". This control, so far as it 

 goes, acts almost perfectly, but it is open to this objection, 

 that as it only corrects the errors of whatever shaft in the 

 clock the contact-wheel is attached to, any error in wheels 

 between that and the telescope screw are unaffected by 

 it ; also I find in practice that when it is attempted to 

 control a clock by alteration of friction, on any heavy 

 quick-moving part, it takes some little time to act, and 



Fig. 2. 



then generally overdoes the correction, causing what is 

 generally termed "hunting." The second form of control 

 is the first which I introduced. 



Fig. I is an elevation, and Fig. 2 a plan, of the arrange- 

 ment which is attached to the back of the main clock- 

 work, and can be seen in Fig. 3 at E, but on too small a 

 scale for description ; A is a portion of one of the uniform 

 clock motion spindles, or any shaft coupled thereto ; 



B, B, B, are the three wheels of an ordinary mitre 

 remoiitoire train driving by weights, w, the scape-wheel, 



C, into the teeth of which gear the pallets, D D, which 

 pallets are driven by the electric pendulum, P. 



The electric pendulum is connected to and driven by 

 a current from any independent clock. To the weight- 



