June 3, 1915] 
NATURE 385 

one made by Sir William Herschel. In a paper com- 
municated by him to the Royal Society in December, 
1781, he reviews the serious difficulties involved in 
determining the parallax of a star by comparing its 
zenith distance at different times of the year. 
Especially there is the uncertainty introduced by the 
refraction of light, and in addition as the angular 
distances of stars from the zenith are changed by 
precession, nutation, and aberration, any errors jin 
the calculated amount of these changes will all affect 
the results. He proposed, therefore, to examine with 
his big telescope the bright stars and see which of 
them had faint stars near them. The bright stars, he 
said, are probably much nearer than the faint stars; 
and if the parallax does not even amount to 1” the 
case is by no means desperate. With a large tele- 
scope of very great perfection it should be possible 
to detect changes in the angular distance of two 
neighbouring stars. By this differential method the 
difficulties inherent in the method of zenith distances 
will be eliminated. Herschel made a great survey to 
find suitable stars, and in this way was led to the 
discovery of double stars—i.e. of pairs of stars which 
are physically connected and revolve around one 
another, just like sun and earth. This was a most 
important discovery, but as the two components of a 
double star are practically at the same distance from 
us they do not serve to determine parallax, for which 
we need one star to serve as a distant mark. 
For another forty years persistent efforts were made 
without success.  Piazzi, in Italy, thought he had 
detected parallax in Sirius and a number of other 
bright stars, but the changes he detected in the zenith 
distances were unquestionably due to errors introduced 
by uncertainty in refraction, or slight changes in the 
position of his instruments in the course of the year. 
Dr. Brinkley, in Dublin, made a gallant effort and 
took the greatest pains. He thought he had succeeded, 
and for many years there was a controversy between 
him and Pond as to whether his results were trust- 
worthy. The state of knowledge of the distances of 
the fixed stars in 1823 is summed up accurately by 
Pond in the Philosophical Transactions :— 
“The History of annual parallax appears to me to 
be this: in proportion as instruments have been im- 
perfect in their construction, they have misled 
observers into the belief of the existence of sensible 
parallax. This has happened in Italy to astronomers 
of the very first reputation. The Dublin instrument 
is superior to any of a similar construction on the 
Continent; and accordingly it shows a much _ less 
parallax than the Italian astronomers imagined they 
had detected. Conceiving that I have established, 
beyond a doubt, that the Greenwich instrument ap- 
proaches still nearer to perfection, I can come to no 
other conclusion than that this is the reason why it dis- 
covers no parallax at all.” ; 
Besides these and other efforts to find parallax in 
the zenith distances of stars, attempts were also made 
to detect changes in the time at which the stars cross 
the meridian, to see if they are slightly before their 
time at one period of the year and slightly after it at 
another. But these, too, were unsuccessful, even in 
the hands of astronomers like Bessel and Struve. 
The best were some observations of circumpolar stars 
made by Struve in Dorpat between 1814 and 1821. 
The following table shows some of the results at 
which he arrived :— 
7 +0:0537 = + 0-075 + 0-034 
x +0-9627'= —0-:136+0-110 
z+ 1-0997' = + 0-175 0-127 
B 7 +0:4027'= +0:305+0-071 
Capella and 8 Drac. ... #+1-1477'=+0-134+0-139 
6 Aurig. and y Drac. ...  7+1-1387'= +0-020+0-117 
NO. 2379, VOL. 95| 
Polaris and e Urs. Maj.... 
e Urs. Maj. and a Cass.... 
¢€ Urs. Maj. and 5 Cass. ... 
Urs. Min. and a Persei 


This table has the merit of not looking wildly im- 
possible in the present state of our knowledge. It 
has the disadvantage of not giving a definite parallax 
to each star. For example, it is impossible to say 
how much of the 0-134” is to be given to Capella 
and how much to 8 Draconis. Further, the probable 
errors, though really small, are nearly as large as 
the quantities determined. 
Struve and Bessel therefore attempted the problem 
by the differential method recommended by Herschel. 
By this time it had become easier to carry out. The 
method of mounting telescopes equatorially had been 
devised, so that the telescope was always kept point- 
ing to the same part of the sky by clockworlk-driven 
mechanism. Struve chose the bright star a Lyre, 
and measured its distance from a faint star about 40” 
away on ninety-six nights between November, 1835, 
and August, 1838. In the focal plane of his telescope 
he had what is called a position micrometer. The 
micrometer contains two parallel spider-threads 
stretched on frames, and the frames are movable by 
screws until the position shown in the diagram is 
reached : the distance apart of the threads is known by 
the readings of the screw-heads. He found that a Lyra 
had a parallax 0-262" with a probable error +0:025". 
Bessel chose the star 61 Cygni as a likely star to 
be near the sun, and therefore to have appreciable 
parallax. 61 Cygni is not nearly so bright as a Lyre, 
but has a very great angular movement or proper- 
motion among the stars. Bessel used an instrument 
called a heliometer. Like Struve’s telescope, it was 
mounted so that it could be driven by clockwork to 
point always at the same star. The object-glass of 
Bessel’s telescope was made by the great optician 
Fraunhofer, with the intention of cutting it in halves. 
Fraunhofer died before the time came to carry out 
this delicate operation, but it was successfully accom- 
plished after his death. 
Delicate mechanism was provided for turning the 
glass, and also for moving the two _ halves, 
the amount of movement being very accu- 
rately measured by screws. Each half gives a 
perfect image of any object which is examined, but 
the two images are shifted by an amount equal to the 
distance one-half of the lens is moved along the other. 
Thus when a bright star and faint star are looked at, 
one-half of the object-glass can be made to give 
images S and s, and tne other half S’ and s’. By 
moving the screw exactly the 1ight amount s’ can be 
made to coincide with S, and the reading of the 
screw gives a measure of the angular distance between 
the two stars. Bessel made observations on ninety- 
eight nights extending from August, 1837, to Sep- 
tember, 1838. The table, taken from a report by 
Main (Mem. R.A.S., vol. xii., p. 29), shows how 
closely the mean of the observations for each month 
accords with the supposition that the star has the 
parallax 0-369” :— 
1837. 
Observed Effect of 
Mean date displace- parallax 
rent o"369 
August 23 +0:197 +0:212 
September 14 + 0-100 + 0-100 
October 12 +0:040 —0:057 
November 22 —0-2T4 —0:258 
December 21 .  —0°322 — 0317 
1838. 
January 14 —0:376 —0:318 
February 5 —0:223 —0:266 
May 14 + 0-245 +0:238 
June 19 + 0-360 + 0-332 
uly 213%. + 0-216 +0-332 
August 19 +0-151 + 0-227 
September 19 -+0:040 +0:073 
