Philosophical Magazine Series 3

ISSN: 1941-5966 (Print) 1941-5974 (Online) Journal homepage: http://www.tandfonline.com/loi/tphm14

II. On Olbers's method of determining the orbits of

comets

Professor Encke

To cite this article: Professor Encke (1835) II. On Olbers's method of determining the orbits of
comets , Philosophical Magazine Series 3, 7:37, 7-25, DOI: 10.1080/14786443508648649

To link to this article: http://dx.doi.org/10.1080/14786443508648649

Published online: 01 Jun 2009.

Submit your article to this journal

Article views: 2

View related articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=3phm20

Download by: [UQ Library] Date: 06 June 2016, At: 13:25

 [7]
II. On Olbers's ~lethod of determining the Orbits of Cornels.
B~ Professor ESrCKr:.*
R. O L B E R S had the kindness to transmit to me last
D year the preceding paper, and to permit me to insert it
in this volume. It completes his excellent method of deter-
mining the orbits of comets in a point in which greater faeility
and a quicker approximation to the truth by trials more me-
thodically conducted had been desired by many of those who
employed this method, and it finishes in some measure the
structure, tile foundation of which Olbers laid in the year
1797.
Dr. Olbers supposes in this addition that his readers are ac-
quainted with his former formulae. Although this method was
Downloaded by [UQ Library] at 13:25 06 June 2016

already carried to such a degree of" perfection in the first me-

moir-I-, that even the master hand of the author of the Theo-
ria motus,&c., made no essential alterations in it, but only some
abbreviations, and although consequently the subjeet might
be considered as exhausted, yet I hope I shall not incur cen-
sure if I once more completely explain this method of Dr.
Olbers. The development of the neeessary formulee will give
rise to some remarks, partly regarding the aeeuracy of the me-
thod in general, partly referring to the above important ad-
dition lately made to it by Olbers himself. Those remarks
are indeed not new, but their being placed together in this
publication will, I trust, by no means be deemed improper.
Lambert's theorem is a main part of Olbers's method. The
manner of solving it given by Olbers admitting of some abbre-
viations, I shall begin with explaining this little improvement.
Denoting two radii vectores of a parabola, the minimum focal
distance of which is q, by r, ~J, the corresponding true anoma-
lies by v, z], and the interval of time between.them by t, the
double area described will be by Kepler's laws,
F=KtCeq,
where K is the constant of the Theoria morns:
K --- 0"0172021.
On the other hand we have

F----fr~dv, fromv = v, to v = v'.

Substituting for r, its value, r = q__.Z___
cos ~ 'o2
* Encke's J'ahrbuch for ] 833, page ~64.
"l" Memoir on the most easy and most convenient Method of Calcu-
lating the Orbit of a Comet, by William Olbers. Weimar 1797.
 8 Prof. Encke on Olbers's M e t h o d o f
we have, by integration,
(1) Kta/2q=2q~(tan{d--tan~v+{tan{dS--{tan{v~).
Introducing instead of the true anomaly and the smallest
distance, the radii vectores and the chord between, we obtain
the important advantage of exterminating the elements of the
parabola from the equation. In applying the method of co-
ordinates it is likewise more convenient to retain distances
only in the formula. W i t h this view put
- v = o f,
and taking k for the chord~ we have
k ~ = r ~ + r ta - - ¢2 r r 1 cos 2f, or -- (r + / ) g -- 4 r r' c o s f ~ ;
consequently
Downloaded by [UQ Library] at 13:25 06 June 2016

2co C4/r,~= +4(r+r'+k)(r+¢--k).

For the sake of brevity we make
(3) r + r t + k = m~
r + r t - - k = n ~,
which is permitted in every case, even with the condition that
m and n shall be always positive; and we have
r + r t = ½ ( m ~ + n ~)
(4) 2cosf C rrl = + mn.
The upper sign for which, f z 90 °, 7] -- v Z 180 °, may be
almost considered as the only one here to be referred to.
By the same quantities s i n f m a y likewise be expressed. For
we have s i n f ~ = sin ½ (d -- v)~
- cos ½v~+ cos ½¢ 2 _ 2 cos ~~r~d -- v) cos ~vcos½v';
and by introducing r and r r we obtain
sinf ~ q q ^ cosf
- - "-~- + r--i - - ~ ~f c , r ri
r + r I -- 2 c o s f v' r ~J

and by substitution from (4),

(5) 2sinfa/rr'-- (m~n) ,/2q,
where, as above, the upper sign is to be taken f o r f z 90 °, and
the lower one f o r f 7 9 0 °.
Separating in (1) the factor contained in the part on the
right-hand side we have,
F = K t v / 2 q = 2q2 (tan{ d - - tan ½v)
(1 + { t a n ,~vt~ + ~tan {t vltan { v + { t a n ½ v~);
and placing in this equation instead of 1, its value deduced
from this equation,
 determining the Orbits o f Comets. 9
cos ½ (~' - v) _ c o s f * / , ' r '
1 +tan½dtan{V=cos½dcos{v q '
and likewise for the one factor its value, viz.
sin { (d -- v) sinf C r d
tan ½ d - - tan ½ v = cos ½ d cos ½ v = ~ '
we obtain
¢
(6) F = K t , / 9 q= 2 r r ' s i n f e o s f + ~ (' d) sinfn,
or after substitution from (4) and (5),
K t , / 2 e = + ½( m ~ n ) m n , / 2 f f + ~ (m -~ n) 3 ,/q,
- */2
Downloaded by [UQ Library] at 13:25 06 June 2016

therefore
2Kt=+(m~n) mn+k(m~-n) s=½(m a~n3),
or
(7) (r + d + k) ¢ ~ (r + r ~ - k) { = 6 K t .
This equation is solved by trials in Olbers's method. Ap-
proximate values of k, r, r t are substituted until the equation
is satisfied. As it is evidently quite indifferent what form is
given to it, we may just as well deduce the value of one of
the variable quantities from it.
The powers being expanded in series, the odd terms will
destroy one another, (having for the moment only regard to
the upper sign,) and we shall have
, 1 3 _~ 1.3.3.5 ,~, _ t,-~&e.
6 K t = 3 k (r +rr) " - ~ / # (r + r t) ~ -- 4.6--~-7i-.10~ ~r-t r )
or

(r + r ' ) ~ - r + r' ¢.6 .r--4--P_ ~\~--~V/

-&c.
k
Reversing this series, so that the value of ~" +-~ is de-
termined fi'om it, and putting for brevity
2Kt
(8) (r + ~J)¢ -- ~' we shall find
k ] 5 ~5 59
r + r ~ - ~ + 2-~ ~8 + ~ 9916 ~7, &c.*
or
2Kt ( 1~, 5 ~4 59 ~6 )
(9) k - - g ( r + e) 1 + + 3-~ +9~ + ....

* Lambert's Indgniorea Orbitw('ometarumProprietates, p. 63.

Third Series. Vol. 7. No. 37. July 1835, C
 10 Prof. Encke on Olbers's Method of
If we consider the intervals of time as small quantities of
the first order, 7 will be of the same order. The factor of
2Kt
~' (r + r I) will, consequently, differ from unity only by quan-
tities of the second and higher orders, and may be conveniently
represented in a table with the argument 7. The calculation
is then very simple. The approximate value of r, r I, ~ being
2Kt
given, C (r + r I) is first calculated, from which 7 is imme-
diately derived, and this gives by the table immediately the
second factor. The comparison of the value of k thus found,
with the one assumed, will prove how far that value is correct.
The value of this factor, which I will designate by/~, so
Downloaded by [UQ Library] at 13:25 06 June 2016

that
5 4 59 76 . . .
(10) ~= 1 +~7 ~+3-~7 +9~
may also be expressed by a finite formula, by which the calcu-
lation of the table is facilitated, and its use immediately ex-
tended to all eases.
k
W e may always put r ~ = sin y, whence

6 Kt __ (1 + sinT)~ -T- (1 -- siny)÷.

(r + ~)~
Assuming, what we may assume, that ~ always / 90 °, we
may use this formula,
(cos ½ T + sin ½ T) ~ -- 1 _+ sin ~,
and the above equation will be transformed thus,
6Kt
(cos } 7 + sin } 7) s ~ (cos ~ ~ -- sin ½T) 8 = (r
~-----~"
+ d)
Taking first the upper sign, we obtain
6Kt
Ocos ~ T~sin ~ T + 2 sin ~ ~ = -(r- - +- - rl)
'~'
to which this form may be given,
(sin } ~ (sin ~ V~ ~ 6Kt
3 \--j~/- ~ \ ~/~ / = e~ (r + r,)÷"
Lambert's formula shows at once, as k Z (r + d), that in
both cases also
6 K t A 2} (r + ~J)}.
Making, therefore,
6Kt
= sin 0,
 determining the Orbits of Comets. 11
we get from the last equation
sin { T = sin ~ 0 */2.
O f the three roots of the cubic equation, one only fulfills the
two conditions, that
s i n ~ 7 / sin 45 ° or z ~ * / 2 ,
and is positive, viz. the one for which 0 has been taken Z 90 °.
F o r the lower sign we have this form:
/cos ~ ~\ /cos ~ T~ 8 6 Kt
S (,---~-) -, ~ */2 ] = ~ (r + ,~)¢'
or cos ~ T = sin ~ 0 */9.
H e r e sin ½ 0 must be Z ~ */2, and at the same time
Downloaded by [UQ Library] at 13:25 06 June 2016

cos ~ 7 7 {- */2. These two conditions are only fulfilled when

0 is between 90 ° and 135 °. It follows froln this that when
the equation
6Kt
- - sin ~,
2¢ (r + rr)¢
gives a value for 0 which is Z 45 °, for the same data only one
solution is possible, for which d -- v Z 180 °. But when
0 7 45 °, there are two solutions, 0 and 180 ° -- 0, the latter of
which answers to d -- v 7 180 °.
T h e value of 0 may be thus written :
S~
sin 0 =
*/8;
and as we have both from the value of sin ~ T as well as from
that of cos ~ T
sin 7 ~- 2~ sin { 0 */cos ~ 0,
and likewise k = (r + r I) sin T,
we have the following complete system of formul0e :
2Kt
= (r+ ~,)---~'
sin 0 =
(tl) ~/8

3 sin } 0 */cos § 0
sin 8
2Kt
*/(r + r') ~''
where
log 2 K = 8"556611'L
By means of these formulae, log/~ has been calculated in the
C~
 12 Prof. Eneke on Olbers's Method of
annexed table for all values of 7, from ~ = 0 to ~ "-- } C 8, in-
creasing by hundredth parts of unity. The logarithms by
which these calculations were made being only correct to 7
decimals, the last figure may be wrong one or at most two
units, which, however, will have the less influence, as Lam-
bert's formula is in this way applied with much greater ac-
curacy than in the common way. In the latter, as k is small
in comparison of r + r t, a small quantity is obtained as the
difference of two greater ones, while there is no such subtrac-
tion in the use of the table. The ease o f ~ " v Z 180 ° has not
been taken into consideration, as the interpolation between the
values of log tk would then be too tedious. Besides, in these
rarer eases there is not the same disadvantage in following the
Downloaded by [UQ Library] at 13:25 06 June 2016

common process.
Lambert's formula shows most clearly how many observa-
tions are required for determining the parabolic orbit. If the
distance of a comet from the earth were known for one obser-
vation, its place in space, consequefitlylikewise its place with re-
gard to the sun, would be given. Every observation which is
used introduces, therefore, one unknown quantity ; for two ob-
servations two radii veetores and the connecting chord k might
be expressed by two unknown distances, but only one equation
between r, r t and k would be derived from them. But three
observations would give three equations for three unknown
quantities, by combining every two of the three radii veetores;
this would be consequently sufficient for solving the problem.
This application would satisfy one of Kepler's laws, viz. that
the times are proportional to the areas described; but it is
quite independent of the second equally important law, that the
comet must move in a plane passing through the sun. This
latter law gives for the three places of the comet an equation of
condition, so that in three complete observations there are four
equations with three unknown quantities. Although, there-
fore, two observations are not sufficient for determining the
parabola, yet three observations are more than enough, and it
will be necessary either not perfectly to satisfy one of the data
contained in the three complete observations, or only to fulfill
one condition resulting from a combination of two data while
the other is entirely neglected.
Olbers's method is founded on the condition that the three
places of the comet shall be in one plane with the sun. It de-
termines from it with a close approximation the ratio of two
distances from the earth, and consequendy also two places of
the comet in space9 expressed by one unknown quantity. This
ratio applied to Lambert's equation between the places and the
time would give, if the substitution were actually made, one
 determining the Orbits of Comets. 1$
equation with one unknown quantity, but of a very high de-
gree, which would have to be solved by trials. It is therefore
preferred to solve the system of those equations by trials.
If we denote by x, Zt, z, the coordinates of a point referred
to the centre of the sun as the beginning of the coordinates,
we have the following, three equations, for three points in the
same plane passing through the centre of the sun:
Ax +By +Cz =0
Ax t +Bff + Cz I--0
A£1+Bffl+ C~t--O.
The equation of condition for these nine coordinates, after
• B C
the rauos- X- and ~ have been eliminated, may be repre-
Downloaded by [UQ Library] at 13:25 06 June 2016

sented in different forms useful for future application. In or-

der to find the factors by which each of these equations must
be muhiplied with a view to eliminate B and C, by which means
A likewise disappears, we have to combine these two equa-
tions :
P 2 / + Pfff + T"ff I = 0
p z + pt zr + p , ~, = O.
W e obtain from them
p proportional to y" £ -- y~z~r

p" # z - ~jz',
or as these latter values are the double areas of the triangles
between the beginning of the coordinates and the points ~ytz~
,y" z", &e., respectively in the plane of the y z, and as these
again are proportional to the areas of the triangles in the plane
of the orbit, these factors become proportional to [rtrrrJ,
-- [r r'], [r rl], if these symbols stand for the double areas of
the three triangles, viz. sun, second and third places of the co-
met; sun, first and third places of the comet; sun, first and
second places of the comets. The negative sign is introduced
in order to avoid negative areas, in as much as all areas are sup-
posed to originate from the turning of the first radius vector
always in the same sense.
Hence the equation of condition has three forms:
[,~ ,~'] x - [, ,~'] £ + [r ,Q x " = 0
(Z2) ['~"3 Z/-- [r~"] S / + [ ~ ' ] # ' - - - - o
[r, ~',] z - [, ~',] ~, + [~ ~,] ~,, = o.
Let us now designate
The three curtate distances of the comet from"~ f.
the earth, by . . . . . . ". . . . . / ~' g"
 1¢ Prof. Encke on Olbers's Method of
The three observed g eocentrie lon g itudes of d, a",
the comet, by . . . . . . . . . . J ~'
The three geocentrie latitudes of the comet, by . ~, ~', ~",
The three longitudes of the sun, by. . . O, O', e",
T h e three distances of the sun from the earth, by R, R', R",
The three times of the observation, by . . . . t, t', t",
Hence we have
x = ~ c o s a - - R cos 0
2 / = ~ sin ~ -- R sin O
z = ~ tan
and the analogous equations for the six other coordinates.
Substituting these expressions in (12) we get,
Downloaded by [UQ Library] at 13:25 06 June 2016

It' r"] (¢ cos a - R cos o ) -- [r r"] (¢ cos a' - R' cos o 3

+ [r r'] (¢' cos a " - R"cos o") = 0
It' r"] (g'.sin a -- R sin O) -- [r r"] (~' sin a' -- R' sin O')
+ [r r'] (~" sin £' - R" sin O") = 0
[ / r " ] f tan 8-- [rr"]g'tanY+ [r r'] e" tan 8" = 0.
In these equations there are five unknown quantities, the
three p, p', p", and two ratios between the triangles. Two
of them, one g and one ratio, may be eliminated. As every one
of these equations separately expresses one and the same con-
dition, and as they are besides independent of the assumed po-
sition of the axis'of abscissae, it will be permitted to simplify
them by changing the position of this axis. The most conve-
nient forms are the following, which result from the former by
changing the position of the axis of abscissae, in the first of them
by O', and in the second, first by a', and again by O~, the
third remaining unchanged.
[r' r" l (~ cos ( a - o')-- R cos ( 0 - - O')) -- [r r"] (~'cos (a'-- O')-- R')
+ Crr'] o')- R"cos(O"- 09) =o
It'r"](~sin (a'--a) + R sin(O--a'))--[rr"] R' sin (0'--£)
(1 S) - - [r r'] (~" sin (d'-- a')-- R"sin (O"-- £)) = 0
[d r t'] ( f sin (a-- 0 t) + R sin (0 ~-- 0)) -- [r r"] ~1sin (a'-- 0')
+ [r r t] (~" sin (d'-- 0 r) -- R" sin ( 0"-- 0')) -- 0
[ d r"] t~tan ~ - I-r r 'r] f' tan Y + [r r'] e" tan ~"= O.
equations which likewise might have been obtained by the
combination of the two first of the preceding ones,
Eliminating ~' from the two last, we obtain
 determining the Orbits of Comets. 15
[~ ~at] g (tan ~ sin (a - - 0') - - tan ~ sin (a' - - Or))
+ [dr"] R t a n St sin (0 t -- O)
+ [r r I] g" (tan Y sin (d'-- 0') -- tan Y' sin (d -- 0'))
- - [r d] R " t a n Y sin (0" -- Or) -- O,
whence
f --_ It' r"] tan ~ sin (~ - o F) - tan ~ sin (~' - - 0 ~)
"~--P']'] " tan ~" sin (o~'-- 0') -- ~-n ~' sinCa"-- O') " P
tan ~' it' r'~ R sin (0' - - O) - - [r r~ R" sin (0" - - 0')
+ ~ " tan Y' sin (a t - - 0') - - tan ~' sin (a" - - 0')
and introducing the analogous symbols
[R t R"] - - R r R" sin (0" - - 0 t)
Downloaded by [UQ Library] at 13:25 06 June 2016

[R R"] = R R" sin ( 0 " - - O)

[R R'] -- R R' sin (0' -- 0),
the second term of the right hand side of the equation may
be written
([[~t r'~ JR' R"]'~ R tanYsin (or -- O)
+ rr' -- [ R ~ ] ] " tan~'tsin(£--O')--tanYsin(a"-O')"
W e have, therefore, assuming
tan ~t sin (a - o*) -- tan $ sin ( £ - 0 t)
Mt
tan ~"sin (£ -- o r) -- tan Y sin (a" - o r) '
tan St sin (o w-- o)
(14) M"=
tan Ytsin (£ -- 0') -- tan Y sin (a" -- 0') '
p, = [r'r"] M, p + ([r'r"] [R_W'B"]~M,,R
[R R'] / '

an expression as yet perfectly rigorous.

The areas of the triangles [r r'], &c., are in themselves, when
the intervals of time are small, little different from the sectors
of the parabola to which they belong, and as little will the
ratio of two adjoining sectors be much different from that
of the two areas of the corresponding triangles, as at any rate
each is smaller than its corresponding sector. For the orbit
of the earth this will be equally the case, and the more so as it
approximates so nearly to a circle, and as there is consequently
a case in which the equality of these ratios may be almost con-
sidered as rigorously true, viz. when the intervals are equal.
But the sectors are proportional to the times. W e may there-
fore suppose as very nearly true,
[rtr t'] [R'R"] t"-t'
[r r'] -- [R R'] = "-~-~--
t"
By these equations the last term in (14) becomes quite
evanescent, and we obtain
 16 Prof. Encke on Olbers's M e t h o d o f
t" -- tt tan ~ sin ( a - - ®') - - t a n 8 sin (a t - - ®l)
(16) M = ~ . tan ~"sin (a'--O') -- tan ~' sin(a"--O i)
p" -- M p,
and have consequently an expression for P ' in given quanti-
P
ties only.
T h e formula (15)contains the only approximate supposi-
tions made in Olbers's method. Their admissibility and the
cases of exception will be investigated afterwards.
It is now required to find, by means of the equations (16),
convenient expressions for r, r", and the corresponding chord
k , which shall contain no other variable but p. For the first
Downloaded by [UQ Library] at 13:25 06 June 2016

two there is no difficulty, as

r~ = x2 +9 • + z ~
~"~ -" x 'w~ -F Zt"~ + z "~.
T h e expression for E, for which we have
k = + +
is more easily found by a simple geometrical construction which
we shall again apply hereafter. Let the place of the sun
in space be denoted by S, the two places of the earth at the
first and third observations by E and E i', those of the comet
by C and C". Let a line be drawn though E" parallel to, on
the same side with, and equally long as, E C. Calling the ter-
mination of the line thus drawn N, and designating its helio-
centric coordinates by x,, y~, z~, we have, by what precedes and
by the construction,
x = pcosa --Rcos®
y -- psin0~ - - R s i n ®
z - ptan~
x" = M p cos ~"-- R" cos O"
(17) 3t r = M e sin ~' -- R # sin ®"
d I= M p tan ~'.
x~ -- p cos ~ -- R " cos ® "
Yl---- p sin ~ -- R" sin ®"
z~ l_ p tan ~.
The two first of these three systems give~ by adding the
squares,
r ~ = 1~~ see ~ -- 2 p R cos (a -- ®) + R ~
(18) r ' ~ = M~g~sec~"~ -- 2 M p R"cos (a"-- ®") + R "~.
Combining the first system with the third, and assuming
x -- x 1 ---- R" cos O" - - R c o s O = g c o s G
(19) y -- y~ = R" sin O" -- R sin O = g sin G~
 determini~g the Orbi/s of Comets. 17
it is clear that g expresses the chord of the earth's orbit
between E and E", and G the longitude of the first place of
the earth when seen from the third. Combining in the same
way the second system with the third, and assuming
x" -- x t = M p cos ~" -- p cos ct = p h cos ~cos H
(20) y"--yj=Mssina"--psina=phcos~sinH
z" -- z, ---- M p tan ~'f -- p tan ~ = p h sin ~,
p h will be the distance of the point N from C", and ~ and H
are tile latitude and longitude of C" when seen from N. As p
disappears in the equations, h, ~, and H are known quantities.
The systems (19) and (20) combined together give,
Downloaded by [UQ Library] at 13:25 06 June 2016

x" - ,r = x " - xI - (x -- xj), &e.

(21) k~=p~h~--2gphcos~cos(G--H) + g~.
The formulae (18) and (21) assume a form more convenient
for logarithmic calculation by representing them as sums of
two squares.
W i t h this view let
cos ~ cos (~ - ®) = cos
(22) cos ~" cos (~"-- ®") = cos q,"
c o s ~ c o s ( G - H ) = c o s ,~,
and we have
r = a J ( ( p s e c ~ - - R cos*)' + R ' s i n 6 2)
(2s) ,.,'= 4 ' ( ( M e scc ~', - rt,, cos , " ) , + rt,,~ sin ~'"9
k = ~/¢((e h -- g cos , ) ' + g~ sin ¢*).
Here ff and q,II are the angular distances of the comet from the
sun, and ¢ is the angle at N in the triangle N C C ~t. Lastly,
the calculation may be still more simplified by introducing a
new variable quantity. I f we suppose
ph-gcos¢=u, or
u + g cos q~
(-24) e = h '

and substitute this value in the first two equations of (28),

every quantity will be expressed by u. As g sin ~ is the per-
pendieular line drawn in the triangle N C C 't fi'om C to the
side N C ~, ~so u is the distance from C ~t of the point of inter-
section of this perpendicular line with the line N C ~t.
The formulae here given, which, it appears, have the form
most convenient for calculation, have been published by Gauss
on the occasion of the second comet of 1813% In the calcu-
• Zaeh's [.German] Monthly Correspondence,vol. xxvlii, p. 804.
Third Series. Vol. 7. No. 37. Juljl 1885. D
 18 ProF. Encke on Olbers's Method of
lation of g, G, h, H, and g, he has adopted a modification
which is easily found, and which, (as all formulae in the whole
paper,) supposes the use of his logarithmic tables, by which,
from the logarithms of two quantifies, the logarithm of their
sum or of their difference is immediately found. Indeed, these
tables facilitate many calculations so much that every practical
calculator should not spare the trouble of getting expert in
using them. I will now place together all the formulm exaetly
in the symbols used by Gauss, (with the sole exception that the
geocentric latitudes are by me denoted by ~,) as all alteration
would be for the worse.
The data of the observations, viz. the three a, ~, O, R, and t,
being given, we have to calculate successively, agreeably to
Downloaded by [UQ Library] at 13:25 06 June 2016

formulm (16), (19), (20), (22), (28), (24),

t'J--t w tan SP sin (a - - ®') - - tan $ sin (£ - - O')
M -- ~ . tan $,t sin (a:'-- o r) - - tan 6' sin (a" o')
R" cos (O"-- O) -- R = g cos (G -- O)
R" sin (O"-- O) ---- g sin ( G - - o )
M - cos ( a " - ~) = cos g cos ( H - a')
sin (a II -- a) - - h cos g sin ( H -- a")
M tan ~,r _ tan = h sin ~1
cos { cos ( G -- H ) cos
cos ~ cos (~ -- ®) -----c o s
(I)
cos ~" cos (d' - o") ----.c o s ~Fw
g sin =A
R sin ----B
R" sin ~ t Stt
. -

h cos 8 =b
h cos Y~

gcos¢--bRcos¢, =c
L g cos ¢ -- b" R" cos gY = c", then

r" = V I I ~.
" [u + e'V )
\ - - ~ r ] + n . B" . f
(ii) /
k = X / ( u u + A A)
and the value of u must now be so determined that
(r + ~' + )~)÷ - (r + r 't - k) ÷ - t" - t
m
 determining the Orbits of Comets. ]9
where
1
m = ~, log m = 0'9862673,

I f we make use of the table given at the end, u must be so

determined that if
2K
= (r + t),
then
2K
k = (r + r")÷ ( t " - t)~,

where log 2 K "- 8"5366114.

Downloaded by [UQ Library] at 13:25 06 June 2016

In making these trials hitherto, a value of u was assumed

and from it r, r u, k were computed. T h e substitution of the
values thus found in Lambert's formula showed whether the
assumed value of u was the true one. I f not, it was varied
until the formula was complied with. But this way of pro-
ceeding has this disadvantage, that in general u, according to
its geometrical signification, does not admit of an approximate
valuation, and one might probably assume at first a value
very far from the truth. T h e process proposed by Olbers
in the preceding paper appears to be much prefbrable.
F r o m practical, and, as far as here applicable, from theo-
retical reasons, it will be perceived that r + C will be sel-
dom or never Z 1, and likewise seldom 7 3. T h e value
r + r " = 2 will. at least in most cases, not be far from the
truth. If, therefore, by means of the table, from r + r 'I -= 2,
the corresponding k is determined, and from this value of k
the u, and~thence r and r", the comparison of the new value of
r + C with the one assumed for this sum, will prove how far
the assure. [)tion was correct. The. new value, may now be again
apphed, and we may thus contmue untd a perfect agreement
is obtained.
T h e formulae being so simple, it is interesting more closely
to investigate the steps of this process. T h e factor/~ may be
considered as a constant quantity, it being at any rate, in all
cases which can occur, little different from unity. Malting,
for brevity, r + r" "-- s, the differential equations of the for-
mul~e, as they are successively applied, will give the following
relations :

d u = __k d k
U
D2
 '2.0 Prof. Encke on Olbers's Method o f
1 M
d g =--~ d u dg" = -ff d u
dr =see~cosCdp
d r u = sec ~" cos C" d ~".
I f we designate the value of s originally assumed by So, its
correction by d So, so that
so + d s o -----s,
or the true value again by s 1, the value of s, resulting from the
assumed value So, and its correction by dsl, so that in like
manner s 1 + d s t -- s = the true value, we have
d st = -- ( P s e e ~ c ° s C +(rP "+s erc'/)~ " c ° s C t ' ) ~ " p-'K~,u d s°'
Downloaded by [UQ Library] at 13:25 06 June 2016

in which formula tile angles C and C II are those angles which

represent the angular distance of the earth from the sun when
seen from the comet, or the angles C and C it, in the triangles
S C E and S C" E". T h e formulasupposes that the assumed
value so is not so fhr distant from the truth as that the differ-
ential for s may not be applied with some approximation;
it is likewise clear, that by d So, d s l, may be understood not
only the corrections of the numerical values but likewise tile
corrections of their logarithms.
On account of the circumstance just mentioned, we may
simplify the formula by putting for the sum of the terms
see ~ cos C and ~ttsee ~" cos C", this one, 2 ~t see ~l cos C I, b.y
which, at least in reference to the general tbrm, nothing is
changed. W e may besides designate tbr the sake of simplicity
pt see ~t, or the true distance, by the single letter ,a t. Besides, in
order to have such quantities only as have a geometrical mean-
ing, let us consider the triangle N C C n above referred to. T h e
sides of this triangle are k, ~ h, g, and the angle opposite to
k is ¢. Let the angles opposite to p h and g be denoted by X
and Z"; we have h .u = kcosxn.ph
= k (k - g cos X)"
By this means we obtain this simple form, if
.4t cos C I k
q= (r+r") " k--geosx
d s I = -- q . dso.
T h e fhctor q can only become negative when either cos C
becomes negative, or g c o s x 7 k . T h e first case supposes
that the comet is within a sphere, the centre of which is the
sun, and the radius the distance of the earth from the sun, so
that it would be very near both to the sun and to the earth.
T h e second case, on the contrary, requires that g 7 k, or as
 determining the Orbits o f Comets. 21
the linear velocity of the comet at the time of the second ob-
servation is ----J (21 the linear velocity of the earth being
assumed = 1, while the ellipticity of its orbit is neglected, so
that approximately
2
17-~, rt72,
consequently s 7 4. Both cases, comets very near to the earth
and the sun~ and comets very distant from the sun, are rare,
especially the latter. Commonly q will have a positive value.
Hence it follows that employing first so, then the following
value s~, next the one resulting from that one s2, and so on, we
Downloaded by [UQ Library] at 13:25 06 June 2016

shall obtain alternately values too small and too great, an ad-
vantage of no small consequence for the rapid approximation.
For the corrections, or that which is to be added to the as-
sumed value in order to obtain the true one, will be
ds I = --qds o
d s 2 = - - q d s ~ = + q2dso
d s~ = - - q d s~ = - - ~ d so, &c.
W e find indeed, in the four examples calculated by Olbers,
this change of signs occurring three different times. Only in
the case of the first comet of 1805, where the case above re-
ferred to occurred, the successive values are so = 2, s 1 = 1"41S,
s~ = l"328, sz = 1"318, which are all in excess.
By means of the same consideration we may, when three
values have been calculated, approximate to the truth by an
easy interpolation. W e will suppose s t, s~, s~ any three suc-
cessive values of s, each of which has been derived from the
one next preceding it. Let us find the arithmetical differences :
s] A Sl
,12 s~
.4 Sa
sa
we shall have approximately
zlsx = (1 + q) d s~
, a s 2 - - - - ( 1 + q) q. ds l
+ q)~dsl,
consequently,
(,a sl) ~
ds 1 ~- A2 $6
( ,a s2) ~
d $ 3 ~-.
A~ s2 '
and generally an approximate value of q,
 22 Prof. Encke on Olbers's Method of
zff s 0
.4 Sl
which, if necessary, might be employed in an insulated trial.
W e may employ d s 1 or d s~ according as the one or the other
is smaller.
The latter selection and in general the more or less rapid
approximation of the trials depend on the value of q. The
first factor, ~r cos C I
r+rrl'
will in almost all cases be a very small fraction. The mag-
nitude of the second factor will, on the contrary, mainly de-
Downloaded by [UQ Library] at 13:25 06 June 2016

pend on the value of X" Agreeably to the construction, X

is the angle formed by the chords of the orbits of the earth and
the comet, if these chords are so placed together that the first
place of the earth coincides with the first place of the comet,
and the directions are made to correspond to the directions
of both bodies. W h e n the motions are parallel, Z will conse-
quently be = 0, and cos X will have its greatest positive value ;
but when the motions are in contrary directions, x -= 180 °,
and cos X has its negative maximum. Hence it follows that
likewise this factor will in most cases, at the discovery of a
comet, be a real fraction, or will at least not differ much from
unity, because by the nature of the case comets will always be
first seen when their course is rather opposed to that of the
earth, and directed towards the earth, than when both move
in the same direction in parallel courses. In the latter case
they might have been seen before. T h e ambiguity which
seems to exist in this manner of instituting the trials, viz. whe-
ther, on account of u~ -- k ~ - - A s, u is to be taken positive or
negative, disappears in practice almost altogether. By the geo-
metrical meaning of u it is clear that it can only become nega-
tive when the angle X 't is obtuse, consequently g 7 k, and, after
what precedes, s 7 4. Cases of this kind are of rare occur-
rence, and may, when they occur, be easily decided.
It appears, therefore, that this way of instituting the trials
has greatly the advantage over the usual process, partly on
account of the value immediately to be applied ofs 0 = ~, partly
on account of the methodical manner in which the successive
assumptions result from the first without any arbitrary suppo-
sitions. Only when u is very small, which quantity for that
very reason will then be derived with less accuracy fi'om the
combination of k and A, the proceeding may perhaps be more
circuitous. The remarks of Dr. Olbers on cases where, in place
ors 0 = 2 immediately, a smaller or larger value may be applied,
 determhdng the Orbits of Comets. 23
are, o f course, likewise here applicable. T h e smallest valueof
s is, agreeably to what is said above, B + B" ; when this value
is itself nearly = 2, in the cases o f comets which are near
the quadrature, or much smaller than 2, for comets which are
nearer to the conjunction or opposition, a greater or less value
m a y be chosen in the beginning, especially with a due re ard
to the values o f b and b" compared to those o f c and d 1. g
I n order to give an application of these formulm I will give
here the calculation o f the second comet o f 1813, which has
become a model b y Gauss's paper, and I will apply it to the
m e t h o d of trials here explained.
T h e observations o f G~ittinger of the 7th, 14th, and 21st o f
April, 1813, gave for this comet the following d a t a :
Downloaded by [UQ Library] at 13:25 06 June 2016

t = 7"55002
t w= 14"54694
t" -- 21"59931.
= 2 7 1 ° 16 t 88 tt ~ --- + 2 9 ° 2 t O"
d -- 266 27 22 Y = + 22 5'2 18
a" = 2 5 6 48 8 ~ " = + 9 53 12
® = 17 47 41 logR --0"00091
0 ' = 24 38 45 log R t = 0.00175
®" = 31 31 25 log R J~= 0.00260.
H e n c e will be f o u n d :
log M = 9"75799
1 3 = 113 ° 4 3 p 57"
l o g g = 9"380~9
= 109 ° 5 t 49 u
~ ' - 44 1S 9
log h ---- 9 ' 8 1 4 7 7
log A = 9"22527
log B = 9"98706
log B" ---- 9'86038
log b = 9"75645
log b" = 0'05028
c = + 0"31365
C" = + 0"95443.
As a first trial (although it m a y be easily seen that s must be
considerably greater than 2) I assume
so = 2 , log so - 0 ' 3 0 1 0 3 .
W e have next, since
t" -- t = 14'04929
log 2 K (t" -- t) ---- 9'68427,
whence the calculation stands t h u s :
 24 Prof. Encke on Olbers's Melhod of
9"68427
0"15051
9"53376 . . . tan ~ = 9 ' 2 S 2 7 3 = 0"1709
log ~ . . . 0 ' 0 0 0 5 3
log k . . . 9"53429.
logA...9'22527 l o g ( u + c ) . 9 " 7 8 6 6 0 logu+c".0"09780
log k . : . 9"53429 l o g b . . . . 9"75645 log b " . . . 0 " 0 5 0 2 8
9"94013 0"03015 0"04752
log u . . . . 9"47442 log B . . . . 9'98706 log B " . . . 9 " 86038
u = + 0"29814 9"86996 9'92349
u + c - - + 0"61179 log r . . . . 0"16019 log r" . . . 0 " 1 2 4 0 3
u+c"= + 1"25257.
Downloaded by [UQ Library] at 13:25 06 June 2016

0"28332
log st 0"44351.
Continuing this value we obtain
log :,, = 0"13612
log 0"10890
log s2 = 0"42375
and by means of this latter again,
log ~ = 0"13933
logr" 0"11092
log s3 ----0"42639.
Forming now the differences thus~
log s 1 = 0%4351
--1976
log s~ ----0"42375 + 2240
l o g s s - 0"42639 + 264
we obtain
(264)~ -- -- 31,
d log sa = 2240
ors log s ---- 0"4~2608s
and by interpolations
log r = 0"13895
log r" -- 0"11068 :
values which are strictly true, as a repetition of the calculation
would prove. Gauss finds 0"13896 and 0"11068. One would
in this example have come near the truth by two trials only~
for fi'om
log so = 0"30103
logs 1 =0'44351 + 14248 - - 16224
- - 1976
log s~ = 0"42375
we get
(1976)3 = + 241s
d logs~-- + 16224
 determining the Orbits of Comets. 25
or log s = 0"42616.
A calculation made with this value would have given the
required vahm with perfect accuracy.
As soon as s and consequently u, p and p" are found, the
elements of the parabola may be derived in various ways
from the two extreme observations. It will, however, be in-
teresting previously to investigate the degree of approxima-
tion obtained by the assumption in (15) and the applicability
of the formulae (16).
[To be continued.]
I l L On the Summation of slowlg converging and diverging
Infinite Series. B# J. R. Yotmo, Professor of Mathematics
Downloaded by [UQ Library] at 13:25 06 June 2016

in Belfast College.
[Continued from vol. vi. p. 354.]
THE foregoing examples have been chosen of the form
S = b - - c x + d x ~ - - e x ~ + &e.
because it is only for slow series of this form that the trans-
formation furnished by the differential theorem offers any
practical facilities. When the series is of the form
b + ex + dx~+ ex + &c. . . . . . . (D.)
we may indeed convert it into
1--xl { b I~-~xAx+ (1-----'~)~A~x
2 - Aax a -'1 ... (E.)

by simply substitutin, g --x for x, in the formula (B.), and this

is a form to whmh the foregoing arithmetical process may be
applied. But it is easy to see that that process would con-
duct us to the original series, and would terminate in the ac-
tual summation of its successive terms; for by the formula
(B.) it appears that the new series, into which (E.) would be
transformed by the process alluded to, would proceed accord-
ing to the powers of" x -+- ( x + 1 ) instead of accord-
1--x Y'~ '
X
ing to the powers of ~ , as at present, that is, it would
proceed according to the powers of x; and, agreeably to this
1
change, the factor F'----x'which multiplies the series (E.), would
become simply 1 ; so that (E.) would thus be converted into
a series of the form
b + d x + d'x ~ + d x ~ + &c.
Third Series. Vol. 7. No. 37. July 1835. E