Theaetetus - Knowledge As Continued Leearning
Theaetetus - Knowledge As Continued Leearning
Theaetetus - Knowledge As Continued Leearning
Brown, Malcolm S.
1 Thomas L. Heath points out that the later commentator Simplicius made remarks
on the style of Eudemus and claimed to be quoting him word for word (x,,x&)d~tv) so that
even at a date later than Proclus, Eudemus' text was accessible. Eutocius speaks of "examin-
ing" the history of Eudemus. (Euclid's Elements, ed. Thomas L. Heath [3 vols., 2nd eel.; New
York: Dover, 1956l, I, 35.)
Eva Sachs, in her definitive work De Theaeteto Mathematico Atheniensi (Berlin, 1914),
finds in Theaetetus Plato's paradigm philosopher: "llle [Theaetetus] re vera philosophus fuit
perfectus" (p. 69). She thinks of him in connection with those "admittedly rare but not im-
possible" Platonic philosophers who are the only guarantee that the intellectual life can
survive in society.
[3593
360 H I S T O R Y OF P H I L O S O P H Y
Within Plato's own thought, and especially in Meno, Phaedo and Republic,
there are close connections ,to the content--mathematical and philosophical---of
Theaetetus. F. M. Corrfford drew up a catalogue of similarities 3 between the
introductory conversation of Meno a n d that of Theaetetus which takes in the
following four items: (1) Meno, like Theaetetus, begins by erroneously offering a
list of answers instead of a single one in response to Socrates' first question; like
Theaetetus, he must be corrected. (2) Socrates asks his respondent to form his
answer upon the example of a mathematical definition. (3) Socrates echoes in
Theaetetus the complaint Meno had entered about Socrates: the complaint about
his powers of reducing others to a perplexity (aporia) like his own. (4) In Meno
the discussion then turns to the process of recollection, as in Theaetetus it turns to
the process of maieutics. One could add to Cornford's list the point Socrates
makes about the priority of the "what" to the "what kind" (the ti to the poion).
At Theaetetus 196D Socrates echoes the point he made repeatedly in Meno (71B,
86D, 100B). These two dialogues are drawn still closer together by the common
topic of incommensurability, which appears explicitly in Theaetetus 147-148 and
is implied in Meno 82-85. Where Meno had dealt by implication (and Republic 546
is to deal explicitly) with V~, Theaetetus carries the problem forward to the cases
of V3, V~, etc.'
Between Theaetetus and Phaedo there is a more fundamental topical connection
in their discussions of Equality/~nequality. A passage in Theaetetus (154-155)
deals with equality "in length or in number" and thus relates to the important
Phaedo discussion (74f) of the Ideas of Equal/Unequal. And in this same passage
the paradox of Socrates' height both remaining the same and "becoming less"
recalls nearly word for word Phaedo 102f. Corrfford's comments to this Theaetetus
passage treat it as continuous with the discussion of "tallness" in Phaedo generally.
The only difference from the discussion in Phaedo is that in the present passage it
is n o t only Greater and Less which are explicitly talked about, but Equal also.
The Theaetetus discussion of incommensurability a n d equality also relates,
a n d in a more consequential way, to a pattern of development 5 in Plato's
epistemology which can be seen extending from Meno through Phaedo and
Republic. In Meno the particular mathematical question of the incommensurability
of side and diagonal of the square is brought into the philosophical discussion by
way of criticism: 6 using it as an example, Plato is able to exhibit a right and a
wrong way to conduct an inquiry. In Phaedo it is the more generalized and more
fundamental mathematical thought about Equality which is represented. In
F. M. Comford, Plato's Theory of Knowledge (New York, 1934), pp. 27-28. Cf.
R. C. Cross, "Logos and Forms in Plato," in Studies in Plato's Metaphysics, ed. R. E. Alien
(New York, 1965), p. 21.
9 Already in antiquity this point of parallelism was made by the anonymous com-
mentator on Theaetetus (eds. H. Diels and W. Schubart, Berliner Klassikertexte ii, 1905):
"the [side of the] two-foot square is also incommensurable . . . but he left it out, they say,
because it is in the Meno" (-~r ~-~ x~[ ~6 ~o~v xt~p~T~vo~ ~tlx~xpov . . . ~ & ~pc~,Os~,
~iv, a6~6 ~tdxtiv x~ MivLovO.
In section V of this paper, I shall return to the question of Plato's development, in
so far as it is reflected in these particular topics.
6 I have argued that the slave-boy's lesson ends in a criticism of sophistical ways of
knowing in an article in Review of Metaphysics, XXI (September 1967), 57-93.
THEAETETUS 361
"'saving the phenomena" for knowledge can be advanced by analogy with the
successful project of saving the irrational for mathematics.
The sense in which Theaetetus ushers in a phase of "self-criticism" in Plato's
dialogues could accordingly be made clearer. For Plato would be recognizing that
his own theory of Ideas, like the Pythagorean theory of numbers, may "prove too
much." That is, Ideas serve to confine knowledge within the (doctrinaire) limita-
tions of a theory of immediate apprehension of intellectual objects---or a glimpsing
of fixed and finished realities. Such a theory of knowing is unresponsive to three
important, and related, facts of intellectual life: (1) sense and opinion are our only
means of access to important kinds of knowledge, (2) in mathematics, there is a
need to "allow for the dynamic" in a whole class of computations, and (3) in
thought generally, there is a need to recognize the centrality of inquiry (anamnetic-
maieuties) so that the fact that even "aporetie" inquiries may be profitable is
taken into accounL In both the mathematical and philosophical cases, one can
proceed from a worse approximation to a better, or, what is equivalent, one can
come to recognize an insufficient answer for what it is, and reject that answer in
favor of a more satisfactory one. Or, to generalize the point, one can find out
(know) that he does not know. The general result of the maieutic efforts of
Socrates and Theaetetus is just this characteristically Socratic "knowing that
they do not know." The general result of Theaetetus' major work in mathematics,
Book X of the Elements, is a sorting and classifying (i.e., a knowing) of irrational
sizes (a'3,oy=), which, according to classical Pythagorean principles of reckoning,
"we do not know." Theaetetus has the same disposition toward knowing mathe-
matical objects as Socrates has toward knowing philosophical ones: Each is un-
conditionally willing to go on inquiring, whether or not there is a guarantee in
advance that the inquiry will come to a successful end----or indeed to any end at all.
Apart from the reference in Plato himself, we have three principal sources
from which to determine the general nature of the mathematical work of
Theavtctus. An anonymous scholiast to Book X of Euclid's Elements says of
theorem 9 of that book,s "this theorem, which is moreover recalled in Plato's
Theaetetus, is the discovery of Theaetetus; but there [in Plato] it is more particular
(?tp~xtSx~pov), whereas here [in Euclid] it is universal." Secondly, in his com-
mentary to Book I of Euclid (Introduction), Proclus mentions the contributions
of Theactetus to the Elements alongside those of Eudoxus: Euclid drew together
many theorems of Eudoxus into his own order and completed many others of
Theaetetus. 9 Proclns' source here, as in the whole of his famous "catalogue of
geometers," is Eudvmus of Rhodes' History of Geometry. The third source, which
10 This text was translated into French b y Woepcke in Mdmoires pr~sentds ~ rAcaddmie
des Sciences, X I V (1856), 658-720.
n "Eudoxos-Studien I: Eine voreudoxische Proportionenlehre und ihre Spuren bei
Aristoteles und Euklid," Quellen und Studien zur Geschichte der Mathematik, B II (1932),
312. B. L. v a n der Waerden endorses the work of Becker in this reconstruction of
Theaetetus' theory and calls attention to the fact that the same result was arrived a t
independently b y b o t h Zeuthen a n d Dijksterhuis (Science Awakening [2nd eeL, New York,
1961], p. 177). ArpAd Szab6's attempt to overturn this consensus (Archiv fiir Begriffs-
geschichte [1964], 151-171) is wholly unconvincing. It rests on a doubtful reading of
the term 6pt~i~6r a n attempt to outmaneuver Alexander, and a simply false assertion that
Tpdt~to never means "prove," still less "prove mathematically" (p. 164). Theaetetus 147D3ff.
seems to have escaped his notice. In this " passage T p ~' t o mamfestly
" means " prove mathe-
matically." Becker had put down the skepticism of Reidemeister and Heath convincingly
with citations from Nichomachus and his commentator Iamblichus, and from three Arabic
commentators on Euclid (Archiv fiir Begriffsgeschichte, IV [1959], 223f.). These citations
give solid support to the connection Becker had argued between antanairesis and both t h e
finite (Book VII) and the infinite (Book X) definitions of proportionality.
364 H I S T O R Y OF P H I L O S O P H Y
II
Plato has both master and pupil, Theodorus and Theaetetus, participate in
the conversation with Socrates, and makes it plain (145C-D) that the younger man
had studied geometry, arithmetic, and the other mathematical sciences under
the older man. In 147D-148B he has Theaetetus outline a set of mathematical
results worked out by his teacher and then tell in some detail how he had
improved upon those results with a theorem of his own. Theodorus had written out
proofs, one at a time, of the irrationality of the "roots" (~uv~tt~) of 3, 5 . . . . 17.
Theaetetus reports that he improved on Theodorus' result by showing how to
divide the whole numbers exhaustively into two classes, one of which would
have only numbers with irrational roots and the other numbers with rational
roots. He also reports the key conception of his generalization: One class contains
"equal-by-equal" numbers and the other "greater-by-less" or "less-by-greater"
(-~ ~),v.'cov ~7='~'~ov~.x~ "~ ~ ) , ~ v ~),~ov~x~;) (148A). In other words, he analyzed
all numbers into those which could be divided into two equal factors and those
whose factors are "always unequal" (~6~,J ~ x=[ [ ) & ~ v ~e[ ~),~up~t) (148A).
It is a controversial question how Theaetetus went about generalizing
Theodorus' results. Heath had suggested a6 that already with Theodorus a new
technique of proving irrationality must have been in force, since if he had
adhered to the traditional Pythagorean line of proof, the generalization of his
theorems about V~, V~, etc., would have been so nearly automatic as to make
both his own effort on individual cases and the latter efforts to generalize the
solution unintelligible. Several historians of mathematics on the Continent, follow-
ing the lead of Zeuthen,:7 developed the idea that Theodorus' new method was
very likely a rudimentary form of the anthyphairesis which Theaetetus was to
develop into a full-blown theory of irrational sizes. Toeptitz ~s endorsed this inter-
pretation of the work of Theodorus, as did Becker, :9 and more recently
van der Waerden has sponsored it. z~ A. Wasserstein, 21 however, has recently
challenged this interpretation of the Continental scholars. Wasserstein argues that
the stumbling block in the way of generalization of Theodorus' work was already
present in the Pythagorean irrationality proof itself. Theodorus would have been
unable, on Wasserstein's view, to reduce cases such as V~ and 1V~ to the simpler
cases of (twice) V~ and (twice) V~, since such a reduction requires the use of
Book VII, theorem 30 and this theorem may have been first added to Book VII by
Theaetetus. Wasserstein foUows Zeuthen in attributing Book VII to Theaetetus.
The general point is that numbers which contain a perfect square factor must be
recognized as analyzable into square and nonsquare factors, so that their square
roots may be reduced to whole number multiples of a simpler square root. For
example, V~ = V4 . 2 = 2 V~. Thus according to Wasserstein Theaetetus, by prov-
ing the theorem which permits the reduction of such cases, first makes possible
the extension of Theodorus' proofs past the case of 17, and indeed to all of the
numbers.
Two considerations make Wasserstein's thesis untenable. First, it is a conse-
quence of Wasserstein's view that van der Waerden must be wrong in his judg-
ment 2z that Book VII of the E l e m e n t s was already in "apple-pie order" before
the end of the vth century. Van der Waerden argues that the elementary theory
of proportion must have been completed during the time of the earlier Pythagoreans
or else Hippocrates of Chios could not have carried out such sophisticated
maneuvers in his arguments concerning the circle and the "lune" around the
mid-vth century. For his arguments depend heavily on proportion theory, and
furthermore it is a striking fact about Hippocrates' arguments that they
scrupulously avoid taking unproved theorems for granted. It is Hippocrates' habit
- - a n d one much admired by the mathematician van der Waerden--to rely only
on theorems which he himself explicitly proves or on propositions which are quite
certainly known (on independent grounds) to have been already part of the "ele-
ments" of his day. In an earlier article 23 van der Waerden had argued, on grounds
both of the internal structure of Book VII and of the use of it by Archytas, that
it must have been complete before his time. In both his article and his book
van der Waerden enforces the view that Book VII, including theorem 30 in it,
slems from a time well before Theaetetus. z4 Thus the need of that theorem could
not have been what held Theodorus back from a more general theory.
III
Theodorus is thought of as (a) a Pythagorean, (b) a geometer, and (c) one who
made the specific contribution of proving the irrationality of l/~, 1 / ~ , . . . Vi-7.
How would Theodorus represent these roots if he were to do so in a way that was
characteristically Pythagorean and geometrical7 He would do so most naturally by
means of Pythagorean (right) triangles whose side and hypotenuse had integral
values with a unit difference, be~nning with (1, 2), (2, 3) . . . . (8, 9). It is an easy
consequence of the so-called Pythagorean theorem that the remaining sides of this
~5 R. Hackforth has shown that ~o~/exo means "got into difficulties," not "stopped."
("Notes on Plato's Theaetetus," Mnemosyne, X [1957], 128.)
368 H I S T O R Y OF P H I L O S O P H Y
series of triangles must have lengths of value V3, Vs, . . . IVY. Theodorus might
naturally have wanted to stop at V~ if he wanted to preserve his Pythagorean
respect for the Decad, and to try to generate all of the dynameis from elementary
numbers properly within the Decad. 26 It is also noteworthy that this series of
triangles begins with the very one with which Theodorus began: the length V~.
Thus if Theodorus was working in a characteristically Pythagorean-geometric
fashion and was drawing upon all and only those pairs of numbers (side,
hypotenuse) which are decadic pairs (i.e., have the values (h--1, h) where h can
have any of the decadic values 2 ... 9), he would have come up with all of the
values V~, ~r~, . . . V~-~ which Plato says he studied--and only those. The more
general point to be noticed is that, if one relaxes the restriction to decadic values,
the sequential integer pairs (1, 2), (2, 3) ... wiU generate, in connection with side
and hypotenuse of a fight triangle, the roots of all odd numbers. This again follows
directly from the Pythagorean theorem: The third side of a triangle whose first
side and hypotenuse are respectively ( h - - l ) and h must have the value
V h 2 - ( h - l ) 2, which simplifies 2~ to hV2-h~-
2 1 , But 2 h - - 1 , where h has values
beginning at 2 (and not stopping at 9), represents a/l odd numbers.
Now there is a pair of theorems which appears early in Book X of the Ele-
ments (the book known to have been written by Theaetetus) which solves the
problem of determining when the difference between a pair of squares will itself
be square. This is equivalent to supplying the general criterion for the rationality of
b2V~-~-a2. Thus in the special case where a and b are successive integers, so that
b2--a 2 will represent all odd numbers be#nning at 3, this criterion of Theaetetus'
serves to distinguish within odd numbers generally (from 3 ... 17 and beyond)
those cases whose roots are irrational. It is a curious point about this criterion of
Theaetetus (theorems 17-18) that it does not serve any other purpose in Book X
or indeed anywhere else in the Elements. It stands as a logical cul-de-sac in that
work. And it is a theorem which cannot have any other purpose, as Heath has
pointed out in his ample note to the theorem, than in application to numerical
problems. But in relation to the problem which we know from Theaetetus 147-148
that he worked on, this criterion would indeed be consequential. It would enable
Theaetetus to decide on the irrationality not just of the handful of odd numbers
whose roots his teacher had examined (3 ... 17), but on all of the odd numbers.
But only the odd numbers are thus sorted, and Plato credits Theaetetus
with a generalization which comprehends all of the natural numbers, not just
=4 Aristotle makes it clear that those who believed that the elementary numbers went
only as far as 10 (g~pt x~ir 8gzd~or excluded the terminal value, 10 itself, from consideration
(Metaphysics 1084A). One of Aristotle's criticisms, in fact, bears on the point that they
(the Pythagoreanizing Platonists) credited to the numbers up to I0 more reality 0z#,~3,d~'~t,~v)
than to 10 itself (ibM.). The general Pythagorean piety towards the Decad is independently
well evidenced in the fragment of Speusippus "On Pythagorean Numbers" (Diels--ICranz,
Fragmente der Vorsokratiker, 7th ed., 44A13).
2T The modern notation may seem to import an alien and anachronistic thought pattern;
the point, however, is very simple and was thoroughly familiar to Pythagoreans even before
Theaetetus. The square on side-2 exceeds the square on side-1 by the odd number 3 (~1~)
and so on up through successive odd numbers, gnomonwise.
THEAETETUS 369
"half" of them. Can the further generalization be accounted for from the known
mathematical work of Theaetetus? Yes it can, and indeed with the help of the
theorems which Theaetetus proves immediately following 17-18 in Book X. The
theorems which follow directly on 17-18, and which culminate in 21-22, investigate
the following question: When is the product of two roots irrational? And the
answer is expressed in the pair of theorems 21-22, namely that the product will
always be irrational when the given roots are not mutually commensurable. Now
this result can be used in a straightforward way to complete the generalization
with which Plato credits Theaetetus. For the numbers remaining to be classified
after the odds are taken care of are obviously all the evens. But the evens may be
subclassified, and were traditionally subclassified, 2s according to the kind of
numbers (odd or even) of which they are products: All even numbers are either
odd-by-even or even-by-even. Let us consider the case of 6. It is the simplest of the
odd-by-even numbers, since it is a product of 3 and 2. The problem of whether or
not the root of 6 is irrational can be analyzed into the problem of whether the
product of the roots of 3 and 2 is irrational. But the irrationality of 1/3 has already
been decided (and similarly for every other odd number) and so has the irration-
ality of ~r~. Furthermore it is a very elementary matter to show that V-3 and 1/2
are incommensurable one with the other, and similarly for the root of every odd
number and ~r~: If the root of any odd number n were commensurable with 1/2,
it would follow that the same number was both odd and even. Therefore this is the
point at which Theaetetus' theorems 21-22 come into play. For one portion of the
numbers remaining to be classified was the portion defined as odd-by-even, and
this portion is now taken care of by analysis of the given number into its odd
and even components, and applying the product theorems 21-22. The only remain-
ing portion of the even numbers, namely the even-by-even numbers, can be
analyzed and handled by means of the product theorems in exactly the same
way. No very serious obstacle to the analysis is presented by the fact that, either
in the case of "pure evens" (expressible as 2") or in the more general sort of
even-by-even, the factor of 2 can be multiplied into itself arbitrarily many times.
For it follows at once from another available theorem (Euclid, IX, 8) that either
2 ~ will be a perfect square or it will be the double of a perfect square 2" 2 "-x.
Thus when it is a question of handling a number in which the factor of 2 appears
several times, it is a routine matter to collect the 2's, and thereby to reduce the
given number either to the double of an odd number (and thus to a case of odd-
by-even) or to the double of a perfect square, or to a perfect square pure and
simple. For every case the foregoing technique of analysis suffices to decide
whether the given number has irrational roots or not. And this completes the
2s Definitions 8 and 9 of Book VII, the book which is the Pythagorean ancestor of
Theaetetns' theory of proportion, define respectively "even-by-even" numbers and "odd-by-
even" according to whether the analysis of the given number into two factors yielded
a pair of even numbers or one even and one odd number. When Aristotle is criticizing
Pythagorean and Platonic speculations about number in Metaphysics XIII, he calls on the
same classification of numbers, according to their "genesis" from factors: (a) the odds,
Co) the pure evens, and (c) the "other evens" (1084a).
370 H I S T O R Y OF P H I L O S O P H Y
generalization of Theodorus' theory, since we have now proceeded from some odd
numbers to all of them, thence to some of the evens (the odd-by-evens), and lastly
to the remaining evens (the even-by-evens). 29
The foregoing interpretation of the Platonic passage in the light of the
beginning of Book X of Euclid offers the first of the two advances I am claiming
to make in the interpretation of this passage. Before going over to the second,
perhaps it is in order to catalogue some results. The foregoing interpretation
achieves the following: (1) it places the theory of Theaetetus in the context of
what is independently attested to be his mathematical work, (2) it makes better
sense of the fact that Theodorus began with 3, 5 ... and ended with 17, and (3) it
establishes a close continuity between the individual proofs of Theodorus and
the generalization Plato credits to Theaetetus. It brings along with it the incidental
advantage of explaining Theaetetus' motivation in theorems 17-18, the criterion
for the irrationality of b2Vb2---a2 which he makes no use of in Book X. Finally, it
enables one to make more sense of the remark of the scholiast to Book X, which
referred to the Platonic p a s s a g e : ~
It remains for me to discuss the second of the ways in which I a m elaborating
the interpretations of the Continental scholars. The key conception which opened
up this line of thinking to Theaetetus was the representation of numbers as
rectangles whose sides might be either equal or unequal. Here again one is
reminded of the mathematics of the Meno, in that Socrates there notified the boy
that he was not seeking an 8-foot (rectangular) area which might be "long this
way, but short that way" (~r x ~ 6 ~ ~ v },~xp6v, ~ ~E ~?~y6) but rather "equal
in both directions, like this [square] one" (~.~,)& ~ov ~v,0t~L,~ ~cto 6 ~ ~oux{)
(83A). The point is that it is not at all difficult to double a given square if no
restrictions are put on the shape of the resulting figure: One would need only
double a given side, while leaving the other side unchanged. But if the doubling
of the size must be done within restrictions as to the shape of the resulting figure
(e.g., if the resulting figure must be equal-sided) the problem is not so easy. A n d
if the further restriction is implied, to the effect that the 8-foot figure must havo
side lengths expressible in terms of the given units, the problem is simply im-
possible to solve:* Using the key conception of Theaetetus in Theaetetus 147-148,
We may say that an area of 8 units is by nature oblong, and therefore cannot be
"made like" a square. No matter how one chooses to transform a rectangle of
8 units within these restrictions (e.g., from 4-by-2 to 8/3-by-3 to 48/17-by-17/6)
he will always get a result which is oblong, never will he "assimilate" the 8-foot
area to square form. On the other hand, if the problem is treated purely gee-
metrically, the assimilation problem is at once solved. Any rectangle can be
"squared," or assimilated to square form 3z geometrically. Thus, putting the two
results together, as it was Theaetetus' main idea to do, the result can be stated:
In the case of numbers, if one thinks of "assimilating" them all to square form, the
resulting squares will in some cases have sides which are not proper lengths (W/=x~)
but are rather virtuals (~uv0~Fr It is the advantage of conceiving the problem
geometrically that, although there are some numbers which, when analyzed into
a pair of factors will "always" be composed of unequals, there are no geometrical
rectangles whose sides must be "always" unequal.
It is a highly significant technical term, 0[~, which Plato has Theaetetus use to
express the fact that some numbers are "always" bounded by greater-and-less
sides (148A). The same technical term occurs in both of the two first propositions
of Book X and introduces the notion of a continued process (of subtraction).
Proclus uses the term to express the continued character of the Pythagorean
"method," which can "always" be carried out a step further. It is also used by
the author of the Epinomis (Plato?) in a mathematical context which led
Toeplitz 33 to agree with Taylor that it may have this same meaning: The meaning
of a process which can be carried out indefinitely f a r ) 4 The process is that of
successively interpolating between a pair of numbers the two means, arithmetic
and harmonic. The point to carrying out such a process is to "contain" the
geometric mean in an ever diminishing interval. For with each step of the con-
tinued process, the interval between the first two means becomes smaller, and the
third mean is always in the interval. But the geometric mean between two numbers
is exactly what is required if the two numbers, taken as a produet~ are to be
which makes very clear the relation of Tbeaetems' work to the problem of "doubling the
square" represente~d in Meno. For the porism presents a generalized solution to that
problem. It provides for what we would call n-tuplicating a given rectilineal area. There is
a faint philological irony in the fact that the same problem which had stood in earlier
Pythagorean times as a paradigm of intellectual aporia is now solved by Theaetetus, along
with indefinitely many related aporiai, by this porism (~ptr162
*= I am berr drawing on the language of the Epinomis. Whether or not Plato wrote
that dialogue, there is presumably a good conformiW between the conception of geometry
and arithmetic in it and Plato's own. V/hat is said (990C) is that "assimilation" (6~to~to~tr
of numbers which are by nature (~6r not similar becomes possible through geometry.
It is also said that this praetematural way of handling numbers is a more than human wonder
88 O. Toeplitz, "Die Mathematische Epinomisstelle," Ouellen und Studien B II, 345.
s4 Toeplitz denies, however, that the infinite process in reference in the Epinomis passage
(991B) is the same as the process of continued fractioning which Taylor claimed to find in
it. Toeplitz showed, in the same article, that Taylor had been hasty in generalizing the
meaning of the mathematics of the Epinomis passage. But Toeplitz does not deny--he evvn
gives further basis for affirming--that a continued process is at issue. Van der Waerden
endorsed this interpretation of the passage and pointed out the intimate connection between
this treatment of the two (arithmetic and harmonic) means and the mathematical work of
Archytas (Hermes, LXXVIII, 185). Archytas says, in a fragment now extant, exactly what
is essential to this continued process, namely that hm~=< GM < am~=.
372 H I S T O R Y OF P H I L O S O P H Y
Each hrn~ and each amk is rational, and as k increases, the values of the paired
means come continually closer to equality. Thus, since the geometric mean (GM)
is always between hmk and amk, the differences between each sort of mean and the
sought-for geometric mean will diminish continually. There is no difficulty in
showing that the interval between the two means "goes to zero" as informal
mathematical parlance puts it nowadays. It can be shown easily with the most
famous of Theaetetus' own theorems: Euclid X, 1. Now in the case of that class
of numbers which Theaetetus defined as having their sides "always" unequal, the
point is that, even though hmk and amk will give ever closer rational approxima-
tions to GM, the value which they are approximating is itself irrational. This can
be expressed in terms of the antanairesis notion as follows: Whereas any given
hmk or amk will have a finite antanairesis with a or b, the GM which they
approach from the side of the lesser and the greater will have the same infinite
antanairesis with either. The main point is that, since one can produce a rational
mean (e.g., a harmonic mean) which will have a ratio to a given extreme
arbitrarily close to that which the geometric mean must have, then the supposition
that the irrational mean does not have the same ratio (i.e., same antanairesis) to
each extreme can be reduced to absurdity. The form of the argument is the
indirect form which later came to be called "exhaustion." The absurdity which
results from the supposition that the irrational has not the same ratio is that the
root must be at once inside a given finite interval and outside it.
The important theoretical point here is that irrational roots of a given number
(which are the same thing as the irrational GM between its pair of factors) can
be handled within the new theory of proportion, whereas in the older Pythagorean
theory of proportion, they were simply impossible to handle. The technical designa-
tion ~,),oT~ expressed the point that they were not comprehensible in ratios 0,6-1,o0
of the classical type. The accomplishment of Theaetetus, then, has both a critical
and a constructive side. The critical point he makes is to decide exactly when
one is confronting a Pythagorean alogon. It is exactly when one is dealing with a
virtuality, not a rational length. The constructive contribution is the new theory of
proportion as "same antanairesis" according to which proportion can now for
the first time handle, by an indirect method, the dynamic as well as the rational.
But there are complications latent in the new theory of proportion and its indirect
method, complications in particular for the conceptions of Equality and measure.
THEAETETUS 373
IV
35 F. M. Comford, p. 41.
a6 In the present passage there is evidence that the paradoxes about greater-equal-less,
reflected both in Phaedo 102 and here, have a basis independent of Plato, and in particular
of his theory of Ideas. For here Theaetetus is said to have puzzled about the point that a
thing of intermediate size must in some sense be paradoxically both great and small--and
to have been puzzled by these things before ever having met Socrates. Extra support is
therefore available from this passage of Theaetems for interpreting the Phaedo discussion
against the background of mathematics. For here it seems to be mathematical speculation
on Equality, proportion, and irrationals which gives rise to the paradoxes. The "dynamic"
character of irrationals excludes any treatment of them in terms of the Pythagorean doctrine
of proportion and Equality. Theaetetus' theory, like Bryson's before him, treats equality
itself as a dynamic, and it is accordingly paradoxical.
s~ What is being compared is not the ratios of the intermediate number 6 itself to the
two extremes, but rather the ratios of the differences ([6--4] and [12--6]) to the two extremes.
Each difference is half of the respective extreme, since 6--4 ----1/2(4) and 12--6 ----I/2(12).
This point will be discussed further in footnote 39.
s~ Bertrand Russell, in his History ol Western Philosophy (New York, 1967, p. 129) tie-
scribes Plato's bewilderment in the following way: "He thinks that if A is greater than B and
less than C, then A is a t o n c e great and small, which seems to him a contradiction. Such troubles
are among the infantile diseases of philosophy." This judgment is as unfair as it is harsh
Erhard Scheibe takes strenuous issue with it in his recent article "13ber Relativbegriffe in tier
Philosophic Platons" (Phronesis XII, 1 [1967], 28-46), mainly on the basis of passages in the
Republic. At least a half-century before Plato the mathematician Hippocrates of Chios
was executing proofs involving the relations of inequality at such a high level of mathe-
matical rigor as to inspire admiration in the xxth century algebraist van der Waerden (Science
Awakening, p. 136). Hippocrates was distinguished not for his "infantilism" but for his
maturity. Van der Waerden notes that Euler, some twenty-two centuries after Hippocrates,
was not up to his level of sophistication about the inequality relations. In the period just
before Plato, and culminating in the work of Plato's own companions Theaetetus and
Eudoxus, the logic of inequality relations had been elaborated and sharpened.
374 HISTORY OF PHILOSOPHY
unless one thinks of the example in connection with Theaetetus' work on ir-
rationals, there simply is no serious mathematical content in this paradox. For
even supposing that there were a philosophically interesting sense in which.one
should say of 6 that it "becomes greater without increasing," it remains true that
there is not any very apparent mathematical interest in saying that. Yet Theaetetus
is represented as having puzzled seriously over paradoxes like those before
meeting Socrates, when his interest in them would have been mathematical.
Let us take a closer look at the text in which the numerical example appears.
Plato does not say simply that 6 is greater than 4 and less than 12, but adds a
point which seems unnecessary to his conclusion, namely that 6 exceeds 4 "'by a
half," and falls short of 12 "being half": ~),etovg z ~ ~e'r~tptov x-~ hF~o),toug..
~8~x= ~),dxxou; x=~ ~Ft~et; (154C). By taking this extra information into account,
and especially accounting for it in the context of Theaetetus' mathematics, we can
get an understanding of what is wonder-provoking and dizzying in the paradox.
For the hemiolioux and the hemiseis express the fact that the difference (2) between
6 and 4 is the same fraction of 4 as the difference (6) between 6 and 12 is of 12,
namely the fraction "half." But when (and only when) this fraction is the same,
a given middle term between two extremes is the harmonic mean in the interval,s9
The identical point is made in Epinomis and indeed in the passage of that work
which has bearing on continued processes and irrationals: "The other [harmonic]
mean exceeds and is exceeded by the two extremes by the same fraction of
themselves" ('r6 ~' ~'r~pov [~.t~ov] x~ =6'~6 F.~ps~"r~v ~xp~v =6":~v 6=~p~Zov "~ x,,i.
sr Theon emphasizes the point that the theory of means is "altogether essential to
reading Plato" (~v='l'xato'~d~ ~ir ":& ID.a~ovtzk ;1 xo6~.tov [IzE~ox~xtov] 8stop~r (p. 113). He then
reviews the kinds of mean, beginning with the arithmetic, geometric and harmonic means,
and brings in the example of the harmonic series 2, 3, 6. This series matches the series 4,
6, 12 of Theaetetus 154(2 perfectly except for the fact that in Theon it is "scaled down"
to half size. Moreover the language in which Theon explains that it is a harmonic series
comes very close to the language of Plato in the Theaetetus passage. Theon says
o'up.~i~'qxr ~-" ~r q:~ &v=~.o'/'~q~ [~:~ ~pFtov'~',~], 'c6v Ft~r 6'pov '~Ep at,,~ p.dp~t XotX,'i' 't:06r ~xpour
~n:r ,.. ~,. x-,;. ';~=tp~X~r o~/ov ~', T', r [2, 3, 6] z=;. Tap 6 ~:~.~ r "c(V "~V-~r ,,~:o~
~Ttcpszst x'~r ":ptd6or xr ~l 6u&r "rffp~=~'~r b~r162 "3"~6 ":~r ":ptd6or (p. 114).Thepoint
is no different when scaled up to the size of Plato's numbers.
40 There is no essential difference between V3, the first of the values Plato ascribes
to Theodorus (147D) and 4V3 which I take to be implied here (154C), since any multiple
of an irrational size is itself irrational. Theaetetus proves this in X, 13 but it is a very
elementary point which cannot have been missed by Theodorus.
THEAETETUS 375
furthermore it is material which, quite unlike the question of the size of the
number 6, is fraught with paradoxes which many mathematicians of Plato's
generation, and especially Theaetetus, had "looked at until they were dizzy." The
first of the three paradoxes Plato presents (155A) is that while "remaining equal
to itself" nothing arithmetical or geometrical should become larger or smaller.
Let us apply this thinking to 4 V~. Theaetetus' theorem tells us that it is not a
length properly so caLled, but rather a dynamis. One way he had of interpreting
this was to say that the sides of any rectangle containing an area of 48 units are
"always unequal." But what does this mean about the side of the square equal
to 48 units? It means that, paradoxically, it must be able to be "larger or smaller"
while still remaining equal to itself. For according to the theory of Theodorus and
Theaetetus, all that can be known about 4V~ is that it is in certain intervals, e.g.,
the sub-interval between hml, aml (6, 8) and also in the sub-sub-interval hm2,
am2 (48/7, 7) and so on continually. But it cannot possibly be known, in a finite
a m b e r of steps, what the exact value of the root in question is. Presumably it has
an exact value, and indeed it is always remaining "equal to itself." Yet with each
new stage in the process of measuring its value it in a certain sense "becomes
greater" (i.e., the lower limit of the interval in which it is found becomes greater)
and also "becomes less" (i.e., the upper limit becomes less). Thus the first paradox
takes on a clear meaning in relation to the geometric mean 4V~ in the interval 4,
12, which it did not have in relation to the harmonic mean, 6. And it is a meaning
which it would likely have had to Theaetetus, who had done mathematical work
on this and related irrational sizes.
The second and third principles also have definite application to the example
paradoxes if they are interpreted in the light of Theaetetus' work on irrationals
and only rather puzzling application otherwise. The second principle states that
something on which neither addition nor subtraction ( ~ e ~po~o,7~o ~
dcpsctpo~'~o)has been performed cannot ever either increase or decrease, but must
be "always equal" (155A).41 The paradoxical thing about the "roots" which
Thcaetctus had dizzied himself studying, however, is that any attempt to measure
them against standard units (and the standard units against them) always resulted
in their "increasing or deereasing" even while having nothing added to or taken
away from them. Bemuse, depending on which stage of the measurement process
one was dealing with, one's measurement value (i.e.,rational approximation) would
9z The proximate context of these paradoxes is Theaetetus' mathematical theory, but the
more ultimate context is Eleatic logic. In the largest fragment of Zeno (B2), he argues from
a principle of additive-subtractive identity (cf. G. Vlastos, "Zeno of Elea" in Encyclopedia
of Philosophy, VIH, 370). The principle is equivalent to the second of these listed here.
Any (non-zero) addendum or subtrahend must change the thing on which it operates. Nor
will there be any change without such operating. The temptation, of course, for Zeno as
well as for a mathematician with a stomach for paradox, is to begin thinking of miniscule
diminutions which might (1) be non-zero but (2) not alter the things operated on. This way
lies dizziness. For an Eleatic to predicate Unity of Being is like a Theaetetan's multiplying
a number by the Unit. In both cases what is operated on is as it were left alone. Similarly
for an Eleatic claim that there is nothing besides Being and a Theaetetan addition (or sub-
traction) of Nothing to a number. Identity is conserved. Zeno's otO~.vof fragment 2, an
identity element for adding and subtracting, is an anticipation of the o68~v in the neo-
Pythagorean Iamblichus, which he abbreviates '0'.
376 HISTORY OF PHILOSOPHY
be less or greater even though the root itself has stayed the same and "always
equal." Thus it is characteristic of irrational sizes that they do violence to this
very reasonable principle about the impossibility of anything's experiencing either
increase or decrease while staying equal. The third principle only reinforces the
other two at a more general level. It simply says that a thing cannot be later what
it previously was not tmless some generation (T~v~r xct~ T~T~,~o-0,,Q has taken
place within it. But to all appearances the "roots" which are in question turn
out later to be something which they were not previously, even though, as mathe-
matical objects, they must be exempt from genesis and generation.
Thus if we bear in mind Theaetetus' own contributions to mathematical
theory, the "puzzles of size and number" can be seen to be part of the seriously
perplexing problem of incorporating irrational sizes into one's theory of sizes,
without resorting to a dizzying (paradoxical) conception of measure and Equality.
Cornford's second complaint, then, that they are not serious perplexities--and
Russell's impatience on this score as well (see above, note 38)----can be answered
this way.
Theaetetus is represented (154C-D) as divided in his own mind about the
merits of the mathematical thinking which led into such dizzying paradoxes.
Socrates impersonates Protagoras or someone like him and asks Theaetetus the
pointed question whether something can become greater or more in any other
way than by increasing. Theaetetus replies both yes and no: that "yes" keeps him
consistent with the replies he has made to the mathematical question about the
intermediate value of the mean, and the "no" keeps him consistent with ordinary
rational thinking. Socrates seems to zet him to admit that he believes more strongly
in consistency with rationality than with the paradoxical thinking about mathe-
matical objects (154D), but then, after he reviews the principles which they
accept about Equality and Inequality and introduces another sample of the
paradoxes involved in making concessions to process in these fundamental notions,
Theaetetus again confesses to his uncertainties: He is sometimes lost in wonder
and "dizzy" when attending to these complicated matters (155C). It is at this point
that Socrates pays him the famous compliment, assuring him that the wonder he
is beset with is a mark of his philosophical temperament.
In order to show the relevance of this discussion of mathematical theory to its
context in the dialogue, let us draw together Plato's theory of sensation and
Theaetetus' theory of continued fa'actioning, and catalogue their common points.
The object sought is (1) an "in-between" (~ts-:,,~6-:Q (154A), which (2) is to be
identified by a process of "measuring and being measured" (=~tz~po61~s~,, . . .
=~y, lx~-cpodpte,~o,~) (15413). Further, (3) the opposite poles of the measure process
are called ~uv&l~etr (156A). 42 (4) The intermediate stages of the process are
"infinite in number, but paired off" (x)dL0st ~t~v ~=et~, 8~%t~t 8~) (156A-B). And
finally (5) the object determined "is nothing in itself, but is becoming for someone
always" ( o ~ v d w t ~v ,,d'c6 x,,~' ~ 8 ~),),~ -:~vt ~ T~yv.~{}a~) (157A-B; el. 153E).
In the case of irrational means the measure process for determining its value has
~z Since the measure process is reciprocal, its terms are symmetrically related, and it
makes no difference whether the middle or the extreme is taken to be the 'dynamic' one.
THEAETETUS 377
stages "infinite in number, but paired." To show "same antanairesis" for any
irrational mean in fact requires just such pairing of infinitely many intermediate
steps. The relevance, then, of the passage about theory of sizes and numbers is
that it permits a refinement of Protagoras' seemingly hopeless doctrine of the fluid
object being "measured" by an equally fluid standard of measure. And more
generally it leads to a hope that, despite the doctrines of Cratylus and the earlier
Plato, the flux of phenomena may after all be "saved" for knowledge. 43
It does not seem exaggerated to say that an aspect of Plato's own uncertainty
and division of mind is here showing itself. The innovations which Theaetetus
introduced in mathematical theory, especially the novel theory of proportion,
require fundamental criticism of Pythagorean theory. The most general feature of
Theaetetus' new conception is that, in order to be able to handle incommensurable
sizes, some theoretical allowances have to be made at the most fundamental level.
A new definition of Equality and measure is required, the essential feature of which
is to replace the notion of an actually complete measurement with a notion which
at least tolerates potential (or dynamic) completeness. This innovation is as
drastic in relation to Pythagorean mathematics as an innovation would be in
Platonic philosophy which conceded that in some cases a piece of knowledge
might be only potentially complete, but still knowledge. This would tend to blur
the distinction between opinion and knowledge. To "make allowances" for the
dynamic in knowledge would thus require a serious self-criticism on Plato's part.
A general review of the dialogue, however, adds plausibility to the view that Plato
was powerfully drawn by the example of Theaetetus, and perhaps also spurred
on by paradoxical problems in his own doctrine of Ideas, to a new stage in his
thinking in which he began to think that knowledge itself may have an unavoidable
dynamism about it.
9s W. V. O. Quine discusses the reasons for various levels of postulated entities "which
round out and simplify our account of the flux of experience" (From a Logical Point of
View, 2nd ed. [Cambridge, Mass., 19~1], "On What There Is," p. 18) in terms of the analogy
of introducing irrational numbers to simplify the laws of arithmetic. Quine identifies Plato
with the (bearded) ontologist who would insist on realities superior to both phenomena of
sense and physical objects with their 'potential' forms. Plato equals McX, in other words.
The interesting thing historically is that the wily Wyman ( = Aristotle) was, or was about
to be, at McX's Academy at the time of the writing of Theaetetus, and that in that dialogue
both phenomenalism and physicalism are "tolerantly" treated. Accordingly both Quine's
counsel of "tolerance and an experimental spirit" (p. 19), and the mathematical analogy
which prompted it in him, seem to have been shared by Plato.
378 H I S T O R Y OF P H I L O S O P H Y
they prove untenable or inaccurate. In this respect Theaetetus is, so to speak, more
like Socrates than Socrates himself. That is, Plato may have found in Theaetetus
a m i n d - - a n d also a sort of knowledge--which allowed him to reinterpret
"Socrates" and make of it a more adequate model of the philosophic mind.
Theaetetus was open to correction. He begins his conversation with Socrates
diffidently, but is willing to venture opinions on the condition that "when I make
a mistake, you will correct me" ( ~ ~t x=~ &F~p~t~, ~ = , ~ 0 p B 6 ~ ) (146C). After his
first definition of knowledge (knowledge is perception) has been fully refuted, he
abandons it and tentatively advances a second (knowledge is true opinion) with
the following expression of readiness to be corrected: "this is my answer; if as we
proceed it no longer seems right as it does now, I shall try to find another" (187B).
When he is about to offer his third and last definition, he presents it as a corrected
version of his second one: He allows that he was misremembering what he had
heard when he recited the second one, but now remembers it in full (true opinion
with a logos) (201C-D). And finally, when even the third of his definitions has
been successfully refuted, he admits that none of his efforts has been adequate
and that he must therefore come away from the conversation without a definite
answer, knowing only that he does not know. There are other, minor indications
of this pervasive feature of Theaetetus' mind, such as his admitting that he is
unclear about an answer, but offering to "answer boldly, and take the risk" of
being mistaken (204B). All of these incidents conform quite precisely to the
formula in which Theaetetus had cast his intention at the be~nning of the con-
versation: "Erring, but being corrected." The most severe "correction" possible
would be the one which wholly eliminated the opinion or estimate and left t h e
inquirer altogether without resources (aporia). But even in that ease he would
come away with that minimum of Socratic wisdom, knowing that he does not
know.
Now the two parts of Theaetetus' mathematical work, as represented both in
Plato's dialogue and elsewhere, bear an essential likeness to the method of
Socrates. It is this likeness which may help to reveal the spirit of self-criticism
on Plato's part which is emerging in this dialogue. The theorem according to
which Theaetetus sorted the numbers into those whose roots were "lengths" and
those which were only dynameis can be quite naturally generalized to the
epistemological level of sorting the possible objects of knowledge 44 into those
which are rationally knowable and those which are not. The sorting itself would
amount to discovering (knowing) that one does not know. But the other side of the
mathematical work of Theaetetus bears a more striking resemblance to the ideal
picture of Socrates. For Theaetetus did not simply identify areas of mathematical
ignorance, but rather developed a procedure for exploring those areas and for
classifying the obscure objects which his procedure allowed him to isolate. Es-
sentially his procedure was to use a measurement process, taken from the field
of "knowable" objects (the so-called Euclidean algorithm from the field of rational
" In Greek there is not nearly the gap between the mathemata as "mathematical objects"
and the mathemata as. "objects of knowledge." Plato tends always to play down this difference
further, but especially in this dialogue, where the discussants begin by eulogizing the young
mathematician for his quickness in learning (mathesis) (144B).
THEAETETUS 379
numbers), and to study the contours of the obscure domain of objects according
to the different ways in which they failed to satisfy the demands of rational
measurement. In this way Theaetetus was able, in Book X of the Elements, to
"take the measure," so to speak, of objects which are not rationally measurable.
The general notion is that, even if a given rational measurement process is
impossible (aporetic) and will not yield a final result, still something may be
learned about the object to which it is applied. For the measurement process, even
if unsuccessful in arriving at a final answer, will nevertheless yield a series of
rational approximations which can be construed as containing "errors which are
being corrected." The epistemological analogue would be a well-directed dis-
cussion of opinions which, even if unsuccessful in arriving at a final answer, would
nevertheless permit of an improvement (even an indefinite improvement) of
opinion. The Theaetetus is itself an example of such a process of correction. 45
Plato may have been of a divided mind about the merits of making concessions
to process. He seems to have been quite unwilling to countenance any such
concessions in his theory of knowledge at the time of the development of his
theory of Ideas (Meno-Phaedo-Republic) even though mathematicians around him
seem to have regularly done analogous things. From the time of Meno onwards,
Plato seems to have been constantly aware of a tendency among mathematicians
to adjust their fundamental conceptions, especially the conception of equality,
to make allowances for ranges of things, especially incommensurables, which
would otherwise be productive of paradoxes. But in Theaetetus, and apparently
in response to a lively sense of the mathematical achievements of this companion
and colleague, Plato seems to be yielding somewhat to an epistemological sugges-
lion derived from Theaetetus' notion of continued processes. This would involve
Plato's thinking that opinion, and perhaps even perception, if they can be
processed in just the right way, ought to be taken seriously. Further, it would
involve his tlainking that knowledge is not fully characterized by the fixed and
finished objects (Ideas) toward which it may proceed, but that it is at least partly
characterized by the approximating process itself. This would mean that at least
in one aspect of it, knowledge is a continued process of learning.
Barnard College