CH. IX.

6]
OTHER TECHNICAL TERMS
137
calIed reductiones ad absurdum. For it is the function of this method
to upset something admitted as clear
1
."
8. Analysis and Synthesis.
It will be seen from the note on Euc!. XIII. 1 that the MSS. of the
Elements contain definitions of A nalysis and Sytzthesis followed by
alternative proofs of XIII. I-5 after that method. The definitions and
alternative proofs are interpolated, but they have great historical
interest because of the possibility that they represent an ancient
method of dealing with these propositions, anterior to Eudid. The
propositions give properties of a line cut "in extreme and mean ratio,"
and theyare preliminary to the construction and comparison of the
five regular solids. N ow Pappus, in the section of his Collectz'ott dealing
with the latter subjecf'l, says that he will give the comparisons between
the five figtues, the pyramid, cube, octahedron, dodecahedron and
icosahedron, which have equal surfaces, " not by means of the so-called
analyticalinquiry, by which sorne of the ancients worked out the proofs,
but by the synthetical method 3 .... " The conjecture of Bretschneider
that the matter interpolated in EucJ. XII!. is a survival of investiga-
tions due to Eudoxus has at first sight much to commend it
4
In the
first place, we are told by Proclus that Eudoxus "greatly added to
the number of the theorems which Plato originated regarding the
sec#on, and employed in them the method of analysis
lS
." It is obvious
that "the section" was sorne, particular section which by the time of
Plato had assumed great importance; and the one section of which
this can safely be said is that which was called the "golden section,"
namely, the divison of a straight line in extreme and mean ratio
which appears in Eucl. n. 11 and is therefore most probably Pytha-
gorean. Secondly, as Cantor points out
S
, Eudoxus was the founder
of the theory of proportions in the form in which we find it in Euclid
v., VI., and it was no doubt through meeting, in the course of his
investigations, with proportions not expressible by whole numbers
that he carne to realise the necessity for a new theory of proportions
which should be applicable to incommensurable as well as commen-
surable magnitudes. The" golden section" would f!Jrnish such a case.
And it is even mentioned by Proclus in this connexion. He is
explaining'l that it is onIy in arithmetic that all quantities bear
raHos (p1}7ot;; 'A"l0") to one another, while in geometry there
are l, irrational" ones (app1J7ot;;) as wel!. ' H Theorems about sections
like those in Euc1id's second Book are common to both [arithmetic
and geometry] except that in which the straight line is cut in extreme
and mean ratio
s
."
1 Produs, p. 255, 8--26.
2 Pappu, v. p .. po sqq. 3 ibid. pp. 410, 17-4I2, 2.
4 Bretschneider, p. 168. See however Heiberg's recent suggestion (ParalipolIlena zu
Euklid in Hermes, XXXVIII., 1903) that the author was Heron. The suggestion is based
on a with the remarks on analysis and synthesis quoted from Heron byan-Nairizi
(ed. Curtze, p. 89) at the beginning oC his commentary on Euc1. Book Il. On the whole,
this suggeston commends itself to me more than that of Bretschneider
6 Proclus, p. 67, 6. 6 Cantor, Gesch. d. Malh. I
s
' p. 141.
7 ProcIus, p. 60, 7-9, 8 ibid. p. 60, 16-19.
INTRODUCTION [CH. IX. 6
The definitions of Ana/ysis and Synthesis interpolated in Eucl.
XIII. are as follows (1 adopt the reading of B and V ~ the only in-
telligible one, for the second).
"Analysis is an assumption of that which is sought as if it were
admitted < and the passage > through its consequences to something
admitted (to be) true.
" Synthesis is an assumption of that which is admitted < and the
passage > through its consequences to the finishing or attainment of
what is sought."
The language is by no means clear and has, at the best, to be
filled out.
Pappus has a fuller accountl :
" The so-called (i,vaAvp,evo,; (' Treasury of Analysis ') is, to put it
shortly, a spedal body of doctrine provided for the use m those who,
after finishing the ordinary Elements, are desirous of acquiring the
power of solvng problems which may be set them involving (the
construction of) lines, and it is useful for this alone. It is the work
of three men, Euclid the author of the Elements, Apollonius of Perga,
and Aristaeus the elder, and proceeds by way of analysis and synthesis.
" Analysis then takes that which is sought as if it were admitted
and passes from it through its successive consequences to something
which is admitted as the result of synthesis: for in analysis we assume
that which is sought as if it were (already) done (ryeryovr;), and we
nquire what it is from which this results, and again what is the ante-
cedent cause of the Iatter, q,nd so on, until ~ b y so retracing our steps
we come upon something already known or belonging to the class of
first principIes, and such -a method we call analysis as being soIution
backwards (dV(L7raALV AVG'tV).
" But in synthesis, reversing the process, we take as already done
that which was last arrived at in the analysis and, by arranging in
their natural order as consequences what were before antecedents,
and successively connecting them one with another, we arrive finally
at the construction of what was sought; and this we call synthesis.
" Now analysis is of two kinds, the one directed to searching for
the truth and caBed theoretieal, the other directed to finding what we
are told to find and called problematieal. (1) In t h ~ theoretieal kind
we assume what is sought as if it were existent and true, after which
we pass through ts successive conseq uences, as if they too were true
and established by virtue of Qur hypothesis, to something admitted :
then (a), if that something admitted is true, that which is sought will
also be true and the proof will correspond in the reverse order to the
at;lalysis, hut (b), if we come upon something admittedly false, that
which is sought will also be falseo (2) In the problematical kind we
assume that which is propounded as if it were known, after which we
pass through its successive consequences, taking them as true, up to
something admitted: if then (a) what is admitted is possible and
obtainable, that is, what mathematicians caH given, what was originally
proposed will also be possible, and the proof will again corresponde in
1 Pappus, VII. pp. 634-6.
CH. IX. 6] OTHER TECHNICAL TERMS 139
reverse order to the analysis, but if (b) we come upon something
admittedly impossible, the problem will also be impossible."
The ancient Analysis has been made the subject of careful studies
by several writers during the last half-century, the mos! complete
being those of HankeI, Duhamel and Zeuthen; others by Ofterdinger
and Cantor should also be mentioned l.
The method is as follows. 1t is required, let us say, to prove that
a certain proposition A is true. We assume as a hypothesis that A
is true and, starting from this \Ve find that, if A is true, a certain
other proposition B is true; if B is true, then C; and so on until
we arrive at a proposition K which is admittedly true. The object
of the method is to enable us to infer, in the reverse order, that, since
K is true, the proposition A originally assumed is true. Now
Aristotle had already made it clear that false hypotheses might lead
to a conclusion which is true. There is therefore a possibility of error
unless a certain precaution is taken. While, for example, B may be a
necessary consequence of A, it may happen that A is not a necessary
consequence of B. Thus, in order that the reverse inference from the
truth of K that A is true may be logically justified, it is necessary
that each step in the chain of inferences should be unconditionally
convertible. As a matter of fact, a very large number of theorems in
elementary geometry are unconditionally convertible, so that in practice
the difficulty in securing that the successive steps shall be convertible
is not so great as might be 5upposed. But care is always necessary.
For example, as Hankel says 2, a proposition may not be un con-
ditionally convertible in the form in which it is gene rally quoted.
Thus the proposition "The vertices of all triangles having a common
base and constant vertical angle lie on a circle" cannot be con verted
into the proposition that "All triangles with common base and vertices
lying on a circle have a constant vertical angle"; for this is only true
if the further conditions are satisfied (1) that the circle passes through
the extremities of the common base and (2) that only that part of the
circle is taken as the locus of the vertices which lies on 01te side of the
base. If these conditions are added, the proposition is unconditionalIy
. convertible. Or again, as Zeuthen remarks
3
, K may be obtained by
a series of inferences in which A or sorne other proposition in the
series is only apparently used; this would be the case e.g. when the
method of modern algebra is being employed and the expressions on
each side of the sign of equality have been inadvertently multiplied
by sorne composite magnitude which is in reality equal to zero.
Although the aboye extract from Pappus does not make it clear
that each step in the chain of argument must be convertible in tl).e
case taken, he almost implies this in the second part of the definition
of Analysis where, instead of speaking of the consequences B, C ...
1 Hankel, Zur GescMchte der Mathematik in A lterthum und Mittdalter, 1874, pp. 137-1 50;
Duhamel, Des mthodes dans les sciences de raisonnement, Part l., 3 ed., Paris, 1885, pp. 39-68 ;
Zeuthen, Geschichte der Mathelllatz"k im Altertum und Mittela/ter, 1896, pp. 92-104;
Ofterdinger, Betriige zur Geschichte der gn"echischen Mathematik, VIm, 1860; Cantor,
Geschichte" der Mathelllatik, 1
3
, pp. 220-2.
2 Hankel, p. 139. a Zeuthen, p. 103.
INTRODUCTION [CH. IX. 6
successively following from A, he suddenly changes the expression
and says that we inquire what it t"s (B)from which Afollows (A being
thus the con sequen ce of B, instead of the reverse), and then what
(viz. C) is the antecedent cause of B; and in practice the Greeks
secured what was wantep by a l w ~ y s insisting'on the analysis being
confirmed ~ y subsequent sYlzthesis, that is, they laboriously worked
backwards the whole way from K to A, reversing the order of the
analysis, which process would undoubtedly bring to light any flaw
which had erept into the argument through the a.ccidental neglect of
-the necessary precautions.
Reductio ad absurdum a variety of analysis.
In the process of analysis starting from the hypothesis that a
proposition A is true and passing through B, C ... as successive con-
sequences we may arrive at a proposition K which, instead of being
admittedly true, is either admittedly false or the contradictory of the
original hypothesis A or of sorne one or more of the propositions B, C ...
intermediatc between A and K. N ow correct inference from a true
proposition cannot lead to a false proposition; and in this case there-
fore we may at once conelude, without any inquiry whether the
various steps in the argument are convertible or not, that the hypo-
thesis A is faIse, for, if it were true, all the consequences correctly
inferred from it would be true and no ineompatibility could arise.
This method of proving that a given hypothesis is false furnishes an
indirect method of proving that a given hypothesis A is true, since we
have only to take the contl'adictory of A and to prove that it is falseo
This is the method of reductio ad absurdum, which is therefore a variety
of analysis. The contradictory of A, or not-A, will general1y inelude
more than one case and, in order to prove its falsity, each of the cases
must l1e separately disposed of: e.g., if it is desired to prove that a
certain" part of a figure is equal to sorne other part, we take separately
the hypotheses < 1) that it is greater, (2) that it is less, and prove
that each of these hypotheses leads to a con el usion either admittedly
false or contradictory to the hypothesis itself or to sorne one of its
consequenees.
Analysis as applied to problems.
It is in relation to problems that the ancient analysis has the
greatest significance, because it was the one general method which
the Greeks used for solving all ,< the more abstruse problems" ('Ta
aCTacf>CT'Tpa 'TWV Trpo/3"ATJp,'T(JJV )1.
We have, let us suppose, to construct a figure satisfying a certain
set of conditions. If we are to proceed at all methodically and not
by mere guesswork, it is first necessary to "analyse" those conditions.
To enable this to be done we must get them clearly in our minds,
which is only possible by assuming aH the eonditions to be aetuaHy
fulfilled, in other words, by supposing the problem solved. Then we
have to transform those conditions, by all the means which practice in
su eh cases has taught us to employ, into other eonditions which are
necessarily fulfilled if the original conditions are, and to continue this
1 Proclus. p. '24'2, 16, 17.
CH. IX. 6] OTHER TECHNICAL TERMS
transforrnation until we at length arrive at conditions which we
are in a position to satisfyI. In other words, we must arrive at
sorne relation which enables us to construct a particular part of
the figure which, it is true, has been hypothetically assumed and
even drawn, but which nevertheless really requires to be found in
order that the problern may be solved. Frorn that rnornent the
particular part of the figure becomes one of the data, and a fresh
relation _ has to be found which enables a fresh part of the figure
to be determined by means of the original data and the new one
together. When tl'lis is done, the second new part of the figure also
belongs to the data; and we proceed in this way until all the parts
of the required figure are found 2. The first part of the analysis
down to the point of discovery of a relation which enables
us to say that a certain new part of the figure not belonging
to the original data is given, Hankel calls the transforma#on; the
second part, in which it is proved that all the rernaining parts of
the figure are "given," he calls the resolution. Then follows the
synthesis, which also consists of two parts, (1) the COllstruction, in
the order in which it has to be actually carried out, and in general
following the course of the second part of the analysis, the resolution;
(2) the demOltstration that the figure obtained does satisfy all the given
conditions, which follows the steps of the first part of the analysis,
the tra11.ifonnation, but in the reverse order. The second part of
the analysis, the resolutioll,- would be rnuch facilitated and shortened
by the existence of a systernatic collection of Data such as Euc1id's
book bearing that title, consisting of propositions proving that, if
in a figure certain parts or relations are give1t, other parts or relations
are also give1l. As regards the first part of the analysis, the tralls-
formation, the usual rule applies that every step in. the chain must
be unconditionally convertible; and any failure to observe this
condition will be brought to light by the subsequent synthesis.
The second part, the resolutitJ1l, can be directly turned into the
COltstruction since that only is givell which can be constructed by
the rneans provided in the Elements.
It would be difficult to find a better illustration of the aboye than
the example chosen by Hankel from Pappus'l.
Givell a circle ABe a1ld two points D, E extental to 'it, to draw
straight tines DB, EB from 'D, E to a puiut B Olt tlle circle such that,
'tI DB, EB produced meet the circle again ll e, A, Ae shall be para/lel
to DE.
Analysis.
S u p p o s ~ the problem solved and the tangent at A drawn, meeting
ED produced in F.
(Part I. Transformati01l.)
Then, since A e is parallel to DE, the angle at e is equal to the
angle eDE. .
But, since FA is a tangent, the angle at e is equal to the angle FAE.
Therefore the angle FAE is equal to the angle eDE, whence A,
B, D, F are concyclic.
1 Zeuthen, p. 93. 2 Hankel, p. 141 3 Pappus, VII. pp. 830-2.