Relational Attributes in Aristotle

Relational Attributes in Aristotle
Author(s): Fabio Morales

Source: Phronesis, Vol. 39, No. 3 (1994), pp. 255-274
Published by: Brill
Stable URL: http://www.jstor.org/stable/4182477
Accessed: 11-04-2018 16:03 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Brill is collaborating with JSTOR to digitize, preserve and extend access to Phronesis
This content downloaded from 192.100.201.151 on Wed, 11 Apr 2018 16:03:30 UTC
All use subject to http://about.jstor.org/terms
Relational Attributes in Aristotle*
FABIO MORALES
Aristotle's theory of relations involves serious difficulties of interpreta-

tion. By attempting to solve some of the problems posed by J.L. Ackrill in
his famous commentary on the Categories (Ackrill, 1963), I hope to con-
tribute to a better understanding of Aristotle's statements on the nature
and status of relational attributes. In general, my procedure has been to
analyse the criteria by which entities are supposed to fall under the cate-
gory of "the relative" (lax 3Qo6; tL). The following topics will be consid-
ered: i) Aristotle's two definitions of relatives in Categories 7, ii) the
pseudo-relational character of the parts of substances, and iii) the three-
fold classification of relatives in Metaphysics A 15. A corollary of these
discussions will be that relations may have played for Aristotle a far more
conspicuous role in the definition of substances and attributes than has
been hitherto acknowledged.
The most extensive treatment of relational attributes by Aristotle is to be

found in Categories 7 and Metaphysics A 15, and the present article is
accordingly centered around these two texts. However, reference will
occasionally be made to passages from the Topics, Sophistical Refutations
and other works, whenever they might shed light on the problems dis-
cussed. Although I am aware that Aristotle's philosophical system experi-
enced important changes during its development and cannot always be
treated as a whole, I have found no evidence that could point to a signif-
icant variation in his theory of relational attributes.
Aristotle avails himself, certainly, of different approaches when speak-
* I would like to thank Dr. Rafael-Tomas Caldera, Dr. Bob Sharples (the Editor of this
journal), and the anonymous referee for their encouragement and valuable comments on
an earlier draft of this paper. I am especially indebted to Prof. Pauline Brachbill de
Marin, who kindly revised my English.
Phronesis 1994. Vol. XXXIX13 (Accepted January 1994)
255
ing of relations. In the Topics, for instance, he uses his conception of rela-
tional attributes,' as well as that of other categories, to analyse the logical
structure of what is said, and to distinguish it from the surface likeness of
the words used for saying it. In the Categories a grammatical criterion, very
similar to one previously used by Plato, is given for the identification of
relational attributes, but it soon reveals itself as insufficient, and is appar-
ently replaced by a more ontological approach. Finally, in Metaphysics A
15 Aristotle presents us with three different classes of relational entities and
a fairly exact description of each of them (as well as with some derivative
uses of the word "relative"), but makes no reference to the problems dis-
cussed in Categories 7. 1 hope to be able to show that all these approaches,
far from contradicting one another, fit into a coherent whole.
It is well known that Aristotle does not have a noun for "relation". Since
he lays emphasis on the "things related", and not on the relation itself, he
uses the technical expression tat JQ6o; TL, "things [said] in relation to
[something else]", in order to address what we might call the "correlates" or
"relata" of a relation (the Greek preposition nrQoo means "relative to", "in
relation to"). But he addresses these relata in so far as they possess relation-
al predicates. Instead of speaking of abstract relations like "slavery" or
"equality", he mentions "the slave" and "the equal".2 As Ackrill observes
(1963, 98), Aristotle does not, of course, mean that the slave Callias or the
number 5 are each a relative (they are a substance and a quantity), but that
they are relatives in so far as they are called "slave" and "equal". Now, I
think everyone will agree that the expression "in so far as they are called"
requires in this context, to say the least, an explanation.
For convenience and clarity of exposition, I will begin my discussion with

Categories 7, and not with the Topics, which is probably an earlier work3
'In this article I shall render the Greek la 7tQo6 TL - depending on the context - by
"relational attributes", "relational entities", or "relatives" (the standard rendering in com-
mentaries). By "correlative" I mean the "converse" relational attribute which corre-
sponds to a given relational attribute, e.g. the correlative of "parent" is "child" (for more
details see note 12). Finally, by "correlates" or "relata" I refer to the underlying entities
that stand to each other in a particular relation, e.g. "10" and "5" are correlates or relata
in the relation of "being the double of' (or "being the half of').
2 One exception is Met. A 15, 102 1b6-8, where Aristotle alludes to "abstract" relations
(without calling them by this name) such as "equality" and "likeness". But this is clearly
a derivative, secondary sense of the word "relative", as is shown by its peripheral
mention near the accidental meanings of the term and its absence in other parts of the
Aristotelian corpus.
' For this claim I base myself on evidence drawn by Kahn (1978, 242 ff.) from the
presentation of the doctrine of categories in the two mentioned works.
256
but contains no systematic treatment of relations. Aristotle opens chapter 7
of the Categories by giving a general definition of relational entities:
(Dl) We call relatives [jrQ6g TLI all such things as are said to be just what they are, of or
than other things, or in some other way in relation to something else. For example,
what is larger is called what it is than something else (it is called larger than
something); and what is double is called what it is of something else (it is called
double of something); similarly with all other such cases. (6a36-b2.)4
"Of' or "than" represent in this translation the Greek genitive, which is

supposed to modify the term standing for the correlate of the relation and
thus serve as an identifying feature for relational attributes. The term desig-
nating the correlate may also appear in the accusative when preceded by the
preposition tQo6, or even in the dative (6b33-36).5 Consider the following
cases: 1) the master is called "master of a slave", 6b29 (borkou, in the
genitive), 2) the mountain is called high "in relation to something else" [e.g.
another mountain], 6b8 (JtQ?S tTQOV, in the accusative), and 3) that which
is similar is called "similar to something else", 6b9 (TlVL, in the dative).6
HI
Let us start by trying to "translate" into modem termninology the way Aris-
totle refers to relations. At first sight, he appears to be taking into account
the following elements:
I a) a relational predicate, standing for a category of Being, i.e. standing for a relational
attribute;7
2a) a subject of which the relational predicate is predicated,8 and
4For quotations from Categories, I follow Ackrill's literal translation (Ackrill, 1963).
The expression 6twoaoiv (Ikkw atb; hl_Qov, "in some other way in relation to
something else" 6a37, in definition Dl above, refers to the two latter possibilities, i.e.
those in which the accusative and dative cases are employed. See Cornford, 1935, 283.
6 Aristotle comes very close to the definition of relational attributes given by Plato in the
Republic and elsewhere; Plato calls relational attributes 6oa tOT! TOLWJta otta Eva(
ToV 438bl (see also 438b4, 199e3, 255c12). After reading an illuminating article by
Erhard Scheibe (1967), I have come to believe that many of Aristotle's distinctions
about relational predicates may have developed while he was still attending the Acade-
my.
7 In the Categories Aristotle is not classifying predicates, but the attributes or properties
these stand for: "The categories classify things, not words" (Ackrill, 1963, 73); "Ari-
stotle never regards predication as a grammatical or linguistic relation of word to word,
nor does he ever speak of subject and predicate as concepts united in judgement" (Kahn,
1978, 248).
8 This need not be primary substance, at least not immediately, as the example given
below will show.
257
3a) a correlate which is linguistically "signalled" by a preposition or a grammatical
case.
Thus, given the statement "10 is the double of 5", "is the double of' would
be the relational predicate, "10" the subject, and "5" the correlate. But, as a
matter of fact, Aristotle never presents the relation this way. Most of the
time, he refers to (pairs of) relatives by saying "the double and the half',
"the double of the half', "the half of the double", etc.,9 with no direct
reference to the things that stand in the relation (correlates, relata). As we
have seen, definition Dl seems to provide a grammatical criterion (3a) to
detect the presence of a relation. But since it is obvious that in Greek many
terms, no matter what category they belong to, can be joined by means of
the preposition nQ6; or the genitive or dative cases, without representing
relational entities, a grammatical case by itself cannot serve as the sole
criterion to detect the presence of a relational attribute.10
I suggest therefore that attention be payed to the specification "just what
they are" (6a36) in definition Dl above. The connection between this "just
what they are" and the appropriate grammatical case could prove helpful in
the search for a criterion to identify relational attributes." But what exactly
does 'just what things are" mean? One answer might be that Aristotle, in
this particular instance, is appealing to language usage, i.e. to the meaning
of the terms involved. This would explain why he always insists on address-
ing the relata through relational terms."2 He seems to be talking about terms
such as "larger", "double", etc., which correspond to relational predicates
("is larger than", "is double of') containing at least two free variables
(subject and one or more objects) which have to be satisfied before the
predicate can provide a complete sentence. In order to determine if this is
actually Aristotle's view on relations, I will next deal with some of the
questions raised by J.L. Ackrill in his commentary on the Categories.
Professor Ackrill points out (1963, 99) that some of the words used by
9 Sometimes a relative term is mentioned separately, but when this happens the existence
of the correlative is tacitly assumed.
'? Similarly, the English expression "John's cow" will not turn "cow" into a relational
term.
" A like emphasis on "just what they are" appears in Plato's description of relational
entities; compare 6oa toTiV okC EdvaCI To Rep. 438dl2; ToiTo 83ueL 9orTv SVmp.
199e3; 60CM tJV W6tV ITQv6; 6X1AXc EIciv cat tt(lv Parm. 133c9 f.
12 Aristotle insists that both relata be adequately rendered - i.e. through relational terms
that elicit their relation - in order that the so called "property of reciprocation" between
a relative and its correlative is seen to hold true; for this "property of reciprocation", that
implies that every relative has a corresponding correlative, or - in contemporary jargon -
that for every dyadic relation there is a converse relation, see 6b28-7a22 and Ackrill,
1964 ad loc.
258
Aristotle to exemplify relational entities (e.g. "slave") are endowed with a
complete sense, i.e. do not need to be supplemented by a correlate, so that
the linguistic criterion of incompleteness would be, to say the least, defi-
cient. Nor do the words "state", "condition", "knowledge" and "perception"
have to be followed in Greek by the genitive case, but they are nevertheless
presented by Aristotle (6b2 f.) as instances of relational entities.
We may choose, instead, to understand the distinctive feature of relation-
al entities as the fact that they stand in a relation to something else. "State"
and "condition" appear at Topics 125a35 ff. as bearing a relation to that
which is the only thing in which they can exist, even as "knowledge" relates
necessarily to some object of knowledge, and "perception" to some object
of perception (125a25-32). But this explanation turns out to be as unsatis-
factory as the previous one. Ackrill, in referring (loc. cit.) to the passages
just quoted, observes correctly that if the actual occurrence of a relation
were all that is needed for counting something as a relative, then almost
everything would turn into a relative, since, for example, according to the
doctrine of the Categories, "every non-substance must be 'in' a substance".
Therefore, neither linguistic usage, nor the fact that an item stands in a
relation to something, seem to qualify as criteria for identifying relational
attributes. In order to throw some light on this point, I will next focus on the
question of the categorial status of the parts of substances.
III
Towards the end of Categories 7 Aristotle mentions the possibility that

some secondary substances, namely the parts of substances, may be mistak-
enly classified as relational, since on some occasions we do call, for in-
stance, a head "someone's head", or a hand "someone's hand". Thus lin-
guistic usage (here: the use of the genitive case) would seem to indicate that
"head" and "hand" are relational entities. In order to avoid this conclusion,
which would endanger the primacy of substances among the categories,
Aristotle gives a new definition of relational attributes:
(D2) Now if the definition of relatives which was given above was adequate, it is either
exceedingly difficult or impossible to reach the solution that no substance is spoken
of as a relative. But if it was not adequate, and if those things are relativefor which
being is the same as being somehow related to something [ots To etvaL tlcbTouv
tT TjQ6g TC Jrw; 9XFeLv], then perhaps some answer may be found. The previ-
ous definition does, indeed, apply to all relatives, yet this - their being called what
they are, of other things - is not what their being relatives is. (8a28-35.)
The interpretation of definition D2, as well as the difference between it and
259
definition D1, have been matter for discussions since antiquity. Especially
obscure in D2 is the meaning of Tlo vctL at line 8a32. While the criterion
contained in definition Dl seems to refer to what is said, what things are
called (secundum dici), the one contained in D2 apparently has to do with
what things are (secundum esse). It has consequently been noticed that the
first criterion is linguistic, whereas the second is ontological. Aristotle
would have been forced to reconsider his first definition, due to the polyse-
my usually present in linguistic expressions.'3 Now, although it seems clear
that the Categories deal with entities, not with words or concepts, most of
the time Aristotle does recur, as an orientation for his classification, to the
linguistic meaning of the terms involved (see Kahn, 1978, 237 et passim).
What else besides language usage do we have for the purpose of identifying
relational entities?
The most promising answer to this question may be found, as Ackrill
suggests (1963, 101 f.), immediately after definition D2, where "the second
criterion is said to have a consequence concerning the necessity of 'know-
ing definitely' that to which something is related (8a35-37), and the fact
that this necessity does not hold in the case of heads and hands is taken to
show that they are not, by the revised criterion, relatives (8bl5-19)." Let us
consider the point in some detail.
Professor Ackrill (loc. cit.) finds the second criterion "too strong", if it is
meant to imply that one could know that A (a given subject) is R (a rela-
tional predicate) only if one knew what it was R of. The criterion is, he
says, not satisfied by relational terms like "half' or "slave": one can know
that 97 is half some other number without knowing definitely what that
number is, and that Callias is a slave without knowing definitely who his
master is.
However convincing this interpretation may appear, one cannot avoid
feeling uneasy about the idea that Aristotle was so careless when he formu-
lated his criterion for the identification of relational attributes. Therefore, I
would like to propose in the following pages an alternative reading of
definitions Dl and D2.
Let us begin with definition DI. I believe it is meant to classify as
relatives those terms whose (true) predication implies, besides the presence
of a subject, the existence of "something else" (i.e. the correlate) which has
to meet certain conditions, and can (but does not have to) be mentioned in
'3 "Die Vieldeutigkeit des sprachlichen Ausdrucks IaBt ein Kriterium eindeutiger Be-
stimmtheit nicht zu. Wenig spater bietet Aristoteles eine neue Definition, die sich auf
das Wesen der Fundamente bezieht. [... I Diese eher ontologische Bestimmung der Rela-
tiva erninnert an Platons Bemerkungen im Phaidon." (Erler, 1992, 580.)
260
the appropriate grammatical case. Since for Aristotle all relatives are con-
crete relatives - as I said, in the Categories he is classifying entities, not
predicates -, a relative generally implies the existence of (at least two)
underlying things (subject and object) that stand in the corresponding rela-
tion.
The criterion contained in definition DI may be described as "linguistic",
because the meaning of the term has to be considered in order to determine
if a relative exists or not. But a commonly ignored - but equally important -
condition for the presence of a relative is the existence of both the (under-
lying) subject (this is a consequence of Aristotle's general theory of predi-
cation) and its specific correlate. In general, no relative can exist without
the simultaneous existence of its correlate (7bI5 ff.).'4 Therefore, according
to definition Dl, the term "slave", when truly predicated, entails: 1) the
existence of the subject, e.g. Callias, 2) the existence of the correlate, i.e. a
given master, and 3) the possibility of eliciting the concrete relation by
mentioning the correlate in the appropriate grammatical case.
I will try to show that Aristotle's criterion for recognizing relational
attributes is based on their peculiar indefiniteness. This indefiniteness may
be eliminated by a "definite knowledge" - and the eventual specification -
of the correlate in the specific instance of the relation. The indefiniteness is
quite evident in relational predicates like "to be the double of', "to be
greater than", "to be similar to", "to be between", etc., which have to be
supplemented by the mention of the intended correlates before they can
build true or false sentences. But it may pass unnoticed in some terms like
"father" or "slave".
The fact that some relational terms like "slave" (Ackrill's counter-exam-
ple) are endowed with a comparatively complete sense does not mean that
they do not satisfy definition Dl; it means that in ordinary language we
often denote or qualify things by alluding to a relation in which they stand
to something else, without mentioning explicitly the latter (we do not esti-
mate this relevant in some contexts). For instance, we say "Callias, the
slave", or "Callias is a slave". Still, even in these cases, we do assume that
the denoted (or qualified) things stand in particular relations to other things.
The reason "the double" or "the greater" do not possess this comparatively
"complete sense" is that they are much too vague to be used in this way.
I believe Plato and Aristotle were no less aware of the indefiniteness of
14 The exception to this rule is represented by relatives like "the knowable", "the percep-
tible", "the measurable", etc., which do not imply the simultaneous existence of their
correlates (cf. 7b22 ff.). Such relatives are called what they are, because something else
(knowledge, perception) can be applied to them.
261
relational predicates than we are, considering their concern over the prob-
lem of a hypothetical denial of the law of non-contradiction. Since Heracli-
tus' time, it had been repeatedly asserted that contrary things, such as "up"
and "down", "pure" and "polluted", were (in some vaguely unspecified
way) identical, i.e. that the corresponding terms could be predicated simul-
taneously of the same subject. This would imply a rejection of the law of
non-contradiction, since it was well established that predicating simultane-
ously contrary terms of the same subject involves a contradiction.'5 (Wheth-
er this was actually Heraclitus' view or not, need not concern us here.)'6
Plato and Aristotle pointed out that contrary attributes can only be predi-
cated of the same subject at the same time if they are understood in a
different respect or in a different relation.'7 The latter differences, however,
often pass unnoticed because of the indefiniteness of the predications in-
volved; if the predicates contain relational terms, their indefiniteness may
be corrected by mentioning the correlates. In most cases the incompatibility
of two contrary relational terms becomes clear once they are referred to the
same correlate: an object can be both big and small, but not big and small in
relation to the same thing; water can be good for fish and noxious for men,
but not good and noxious for men (in the same sense). An analogous clause
applies to contradictions: the impossibility of affirming and denying the
same attribute from the same subject holds true only if the attribute is both
affirmed and denied in the same respect and - eventually - in relation to the
same thing.'8
There is textual evidence supporting the assumption that Aristotle regard-
ed relational terms as indefinite. In a passage from Metaphysics N I it is
maintained that, of all types of categorial entities, relatives are those which
"are least of all a substance and a real thing" (1088a29-bl); the reason
given is that a thing may be said to be "less", "greater" or "equal" without
having to change, only because that with which it is being compared has
'" See, for example, Aristotle, Met. F 6, 101 I bl7.

16 "Die eigentliche Erklarung dafiir, daB in der antiken Philosophie zunachst die Relativa
auftraten, die Relation aber unbekannt blieb, muB selbstverstandlich eine historische
sein, und sie ist zu geben aufgrund des allmahlichen Hervorgehens der Relativa aus den
Gegensatzen, wobei in diesem Vorgang Platon eine Mittelstellung zukommt zwischen
Heraklit einerseits und Aristoteles andererseits." (Scheibe, 1967, 45.)
'7 Thus Plato, Rep. 436e7-10: "No such remarks then will disconcert us or any whit the
more make us believe that it is ever possible for the same thing at the same time in the
same respect and the same relation to suffer, be or do opposites." (Translation by Paul
Shorey, my italics.)
18 Aristotle, Met. 1005b19 f.: "The same attribute cannot at the same time belong and not
belong to the same object in the same respect [xaT6 To aw'T]r".
262
changed. This shows clearly that the correlate is not included by Aristotle in
the relational attribute, and consequently, that he understands relatives as
indefinite.
The contention that the knowledge of a relative involves the definite
knowledge of the correlate in that particular instance of the relation (8a35-
37) confronts us with a dilemma. For if the relative as such has to include
the mention of the correlate, then it is difficult to see how it can be indefi-
nite at all. On the other hand, if the correlate does not form part of the
relative, why is it impossible to know the latter without knowing the for-
mer?
Now, since for Aristotle all true relatives are concrete instances of rela-
tives, then, in order to ascertain if a relative exists, one has to examine not
only the subject to which the relative is supposed to apply, but also the
things with which this subject is supposed to relate: for the purpose of
determining if a man is a "slave", or a number a "double", one has to
examine the way in which the man or the number in question is related to
other things. And the other way round: if we know with certainty that a
slave or a double exists, then we know by the same token that some corre-
sponding master or half exists.
As we saw, Ackrill challenges (1963, 102) the contention that a definite
knowledge of the correlate is needed for the definite knowledge of a given
relative. One can know that Callias is someone's slave, he says, without
knowing definitely whose slave he is, or that the number 97 is half some
other number without knowing definitely what this other number is. But in
each of these counter-examples we are assuming that a relative exists (e.g.
because someone has told us, or because we are acquainted with a general
mathematical law); we are not inferring its existence from the basic ele-
ments that constitute the relation.
Aristotle, on the contrary, is trying to specify the necessary and sufficient
conditions for a relative to exist (to be). This is shown by the example he
gives in the passage under discussion (8b7-15): We cannot affirm, he con-
tends, that a given thing [e.g. a rose] is "more beautiful" just by examining
it; we must also determine if there exists something which is "less beauti-
ful".
The extent of the demanded knowledge of the correlate has often been
discussed in commentaries. Aristotle keeps silent on this point, but I think it
is evident that the knowledge required will vary from case to case, depend-
ing on the meaning of the relational terrn involved. It is the relational term's
meaning which tells us what to look for in eventual correlates. In order for
something to count as a correlate, 1) it must exist, and 2) it must satisfy
certain conditions implied by the meaning of the relational term (in our
263
example: it must meet the criteria for something to count as "less beautiful"
than the rose).
Now, why does Aristotle contend that the parts of substance do not
satisfy the aforesaid requirement, that is, how can we "definitely know"
them without knowing definitely the things to which they belong (8bl5-
18)? The question is all the more pressing because, as Ackrill correctly
remarks (1963, 103), Aristotle does not regard a severed hand or head,
strictly speaking, as a hand or head at all.'9 And this is due to the fact that
"all things are defined by their function and capacity, so that when they are
no longer such they should not be said to be the same things, but homonym-
ous" (Politics 1253b20-25). Accordingly, the parts of substances would
seem to be, of all entities, those which most require that reference be made
to the wholes to which they belong.
Paradoxically, I think it is precisely the latter quotation which gives us a
clue as to why the parts of substances do not meet the second criterion and
thus do not fall under the category of Tct TQ6og TL. For since one must appeal
to the whole in order to define them, this reference is contained within the
definition (and the meaning!) of the respective terms. They have to be
considered, at least in this respect, definite enough. It is thus unnecessary -
indeed, redundant - to refer the terns which denote parts to the wholes of
these parts. In the notion of "hand" or "head" it is already implied that these
parts fulfil a particular set of functions in the living organism. Let us as-
sume for a moment that the definition of "hand" were "the part of the body
that enables a living being to seize things to his advantage": then "part" will
obviously be a relational term requiring further specification, but "hand"
will not. The parts of substance cannot be considered as relational, because
the free variable of their genus ("part") has been already satisfied. To deter-
mine "hand" as "hand of a body", or "sight" as the "sight of colours",
would be "to say the same thing twice".'o
There is a passage which strongly suggests that the reference to a corre-
late is understood as a lack of definiteness, and that such a reference be-
comes unnecessary once it has been included in the definition of the corre-
sponding term:
[All arguments of this kind occur] also in dealing with any terms which, though
they are not relational terms at all, yet have their substance, viz. the things of which
they are the states or affections or what not, indicated as well in their definition,
"1 "[...] a severed finger is a finger only in name", Met. 1035b23-25.

20 In a previous version of this paper I had tried to draw support for this thesis - which I
still believe correct - from Met. A 15, 1021 a3 1 -b3. I owe to my anonymous referee
awareness of the fact that the latter text will not bear my previous interpretation of it.
264
they being predicated of these things. Thus e.g. odd is a number containing a
middle; but there are odd numbers; therefore there are numbers numbers containing
a middle. Also, if snubness is a concavity of the nose, and there are snub noses,
there are concave noses noses. (Sophistical Refutations 173b5- 11.)2
At first sight, this passage seems to contain just one more of the many
pieces of advice given for the avoidance of logical mistakes in dialectical
contests. But, as we shall see below, the inclusion of "snubness" as an
example in the discussion in Met. Z 5 about the kind of definability proper
to nonsubstantial entities, with its insistence on the danger of "saying things
twice", speaks in favour of the metaphysical significance of the aforesaid
distinction.
When Aristotle says in Categories 8 that "grammar" is a quality, while its
genus "knowledge" is not, and appeals thereby to linguistic usage (i.e., we
do say "knowledge of something", but not "grammar of something"), the
linguistic mark is meant as an indication, but nothing more. The decisive
factor is that each of the "knowledges" is determined by the realm of reality
to which knowledge is respectively applied. Thus "grammar", a particular
knowledge, does not have to (although it can) be explicitly put in relation to
its subject-matter (i.e. "letters"). But according to Topics 142b31 it needs to
be defined as "knowledge of [writing and reading] letters". While "knowl-
edge" is an indefinite, relational attribute, "grammar", its species, is not.22
(The Topics contain more than one example of non-relational entities which
are defined by means of a relational genus, and Aristotle explicitly allows at
124bl7-22 for a genus to be a relative while its species is not.)
Now, since "grammar" and the rest of the "knowledges" are qualities, but
"knowledge" as such is a relative, we face the difficulty of a single item,
"knowledge", falling under more than one category. (See Aristotle's con-
tention at Topics 121 a5-9, that the genus has to belong to the same category
as the species.) At Categories 8, 1 1a37 we read that there is nothing wrong
with a thing being counted in two genera, an assertion Ackrill (1963, 108)
justifiably finds perplexing.23
21 Translation by W.A. Pickard-Cambridge (Barnes, 1984, I), my italics.

22 Trendelenburg (1846, 126) was already conscious of this "definitional procedure" in
Aristotle, but he did not make, as far as I know, any attempt to explain it: "Hingegen,
wenn sich der Begriff der tJTLGT1[L1 durch die Aufnahme des Gegenstandes zur be-
sonderen Art fortgebildet hat, so befriedigt er sich grammatisch in sich und wird daher
unter die Qualitat gestellt, z.B. YQaIaILRCLX (Cat. 8, 1 1a20, Top. IV 4, 124bl5)."
23 Ackrill's attempt to find a solution to this apparent inconsistency by recurring to a
different classification of the sciences at Topics 145al3-18 (thus suggesting that "gram-
mar" is no true species of "knowledge") overlooks the passage at 124bl7-22, where
"grammar" is unmistakably presented as an instance of a non-relational species that falls
under a relational genus ("knowledge").
265
Obviously, two different classifying criteria for "knowledge" are at stake
here, and I believe this represents a problem for the unity of Aristotle's
categorial framework, a problem which cannot be discussed within the lim-
its of this article. So much, though, seems clear to me: the reason for
"knowledge" being classified as a relative is its indefiniteness. Just because
there is no "abstract knowledge", that is, knowledge devoid of some defi-
nite object of knowledge, "knowledge" as such is indefinite.
Now we are in a better position to interpret definition D2. Since the parts
of substances admittedly satisfy definition Dl (that is, the meaning of the
terms used to designate the parts of substances imply, when they are truly
predicated, the existence of their corresponding wholes), and would have,
therefore, to be counted as relatives, Aristotle sees himself forced to in-
troduce a second definition. Now, if I am right in assuming that all relation-
al attributes are for Aristotle essentially indefinite, then it is easy to see why
the parts of substances do not meet the criterion contained in definition D2,
i.e. do not have to be considered as relatives. They are definite, so their
"being" (to ElvaL, 8a32) - and that means in this context: their categorial
status - is not "the same as being somehow related to something".
It may perhaps be objected that if a relational genus can be so determnined
that it tums into a non-relational item, then this possibility should be grant-
ed to all relatives. In this way, every relational predicate would have a
chance to be supplemented by means of the definite knowledge of its corre-
late and, once being thus "satisfied", turn into a non-relational predicate.
This seems to be a serious objection, for which I have no reply. I will
suggest in section IV, though, that Aristotle was probably aware of an
important distinction between those relatives that can be replaced by non-
relational entities and those that cannot. Perhaps it is simply rooted in the
nature of things that some relations form an essential part of substances or
attributes, while others are just accidental. In cases where the relation is
accidental to the underlying entity, the correlate remains, so to say, just the
contingent individual it is, and external to the relation; but if the relation is
essential, some universal aspect of the correlate will be allowed to "fuse"
with the relational predicate, the predicate itself turning into a complete
notion and coming to form part of the definition of the corresponding item.
IV
In order to make the mentioned distinction between two fundamentally

different types of relatives clearer, I shall next consider the classification of
relatives which is given in Metaphysics A 15 (leaving aside, for reasons of
266
space, some secondary meanings of the word "relative" recorded towards
the end of this chapter). Aristotle proceeds, as usual, by distinguishing the
various ways in which things are called "relatives" (QQ6; nT). He does not
specify if the corresponding items stand for different classes of the same
category, or for generically different things (i.e. totally different meanings
of the word "relative"). Only relatives that correspond to dyadic relations
are dealt with, and spatial relations, among others, are omitted without any
comment. This notwithstanding, the chapter certainly supplements in many
ways our picture of relational attributes as seen by Aristotle.
Three main groups of relatives are distinguished in Metaphysics A 15: 1)
those which are said to be "according to number" (xczt' &QLf}t6v), 2) those
which are said to be "according to a capacity" (xatxait b3vcvaxLw), and 3)
those which are said to be "as the measurable to the measure" (.b; To
[tE*TQflTOV JTQO;T TO tQov). Let us consider them separately.
1) Relatives said to be "according to number" (xaT' &QL6ov). Some
illustrations are: the double to the half, the treble to the third, that-which-
exceeds to that-which-is-exceeded, etc.24 It can be inferred from Aristotle's
indications that all these relational attributes result whenever two things are
compared by means of some quantity (either definite or indefinite). Thus, in
the pair of relatives "that-which-exceeds and that-which-is-exceeded" the
standard of comparison is the size of the smaller object, which is said to be
(indefinitely) surpassed by the size of the bigger one. Now, since Aristotle
usually gives to the word "the one" the meaning of "standard of compari-
son" ([FiTQOV 102 1al 3) - beyond its narrower sense of quantitative unity -,
and since "the one" is the principle of number, he goes on to include in this
first group those relational terms in which the standard of comparison is not
a quantity but a quality or an essence, terms such as "the similar" and "the
same".25 Thus something is called "equal" to something because both things
have a common quantity; something is called "similar" to something be-
cause both things have a common quality; and something is called "the
same" as something because both things have a common essence (1021al0-
13). In these three relations, a particular quantity, quality or essence has to
be taken as "the measure of comparison". Moreover, it is natural to assume
that "the dissimilar", "the diverse" and "the unequal" belong also to this
group, since they imply, no less than their opposites already mentioned, one
24 In these and the following examples the words respectively connected by the hyphen
correspond to one single term in Greek.
25 Since equality, sameness and similarity are symmetrical relations, Aristotle does not
bother to mention both the relative and its correlative (e.g. "the equal and the equal"), as
he does with other relations in the preceding and the following examples in Met. A 15.
267
common standard of comparison. For instance, two things are called "dis-
similar" because one of them has a quality that the other is lacking.26 And it
is worth noticing that most of the examples of quantitative relations careful-
ly distinguished at the beginning of this passage are in fact specifications of
inequality.27
2) Relatives said to be "according to a capacity" (xa'rit 66v%ttv). Some

examples are: that-which-can-heat to that-which-can-be-heated, that-which-
is-heating to that-which-is-being-heated. Aristotle refers generically to this
group as "the active to the passive". In our own words, we may say that it
contains those relatives which are called what they are because something
acts, has acted, or can act, on something else (and therefore, because some-
thing is being acted upon by something else, etc.). A cause-effect relation is
involved in this kind of relatives. Thus "the father and the son" fall under
this group, because at some time the father must have been the cause for the
existence of the son (1021a23 f.). It seems natural to understand "the master
and the slave" (an example from the Categories 6b30, 7a34 ff.) as a specifi-
cation of the relatives "the owner and the property",28 which in turn would
undoubtedly belong here, since for the latter relation to exist some event
justifying the property (acquisition, war, inheritance, etc.) must have taken
place, an event which implies a cause-effect relation.
3) Relatives said to be "as the measurable to the measure" (); T6o
>ETQrlToV T[QOg To [LtTQOV). This group of relatives includes some pairs
previously named in the Categories, such as "knowledge and the knowable"
and "perception and the perceptible". It is described in contrast with the
first two groups:
Relative terms which imply number or capacity, therefore, are all relative because
their very essence includes in its nature a reference to something else, not because
something else is related to it; but that which is measurable or knowable or think-
able is called relative because something else is related to it. For the thinkable
implies that there is thought of it, but the thought is not relative to that of which it
is the thought; for we should then have said the same thing twice. (1021 a26-33.)29
26 At Top. 125al ff. "the contrary", "the different" and "the diverse" are introduced as
relatives ordered according to a species-genus hierarchy. From other passages it is clear
that Aristotle considers "the contrary" as a species of "the different" ("contrariety" is
"the greatest difference", 1055a5), and "the different" as a species of "the diverse"
("difference" is "a diversity in the same genus", 1054b23 ff.).
27 For a superb classification of Aristotle's "quantitative" relatives, see Ross, 1924, 1, ad
loc. A genus-species arrangement of some relational attributes classified in Met. A 15 as
belonging to the first group is given at Topics 121a4 ff.
28 At Cat. 8a24 "property" (xT[Lta) is treated as a relational attribute.
29 Ross' translation (Barnes, 1984, II).
268
In Categories 7 Aristotle had already maintained that most pairs of relatives
are "simultaneous by nature", which means that, for each pair, the de-
struction of either one of the relatives carries the other to destruction: thus,
if the master disappears, the man who was his slave ceases to be his slave; if
a particular instance of double does not exist, neither does its corresponding
half, etc. (7bl5-22). But some relatives, such as knowledge and the know-
able, perception and the perceptible, were said to be exempted from this
rule (7b22-8al2): the destruction of knowledge does not produce the de-
struction of the knowable, nor the destruction of perception the disappear-
ance of the perceptible. The knowable would keep existing as such, even if
man should disappear from the face of the earth. One is thus inclined to say
that this group distinguishes itself by the existential autonomy of one of the
relata in each pair of relatives.
Sir David Ross sums up his commentary on the above quoted passage
from Met. A 15 as follows: "At the bottom of Aristotle's thought, though
not very satisfactorily expressed, is the conviction that knowledge and per-
ception are relative to reality in a way in which reality is not relative to
them." (Ross, 1924, I, 331.)
But besides the existential autonomy of one of the relata in each pair of
relatives, another aspect seems to be distinctive of this class of relatives,
namely, logical priority. I mean the fact, which has been previously men-
tioned, that in some pairs of relational attributes one of the relata may be
used to determine the other. For example, the knowable (i.e. the object of
knowledge), being prior to knowledge, gives knowledge its specific content
and may be used to define it: grammar may be defined as (a kind of)
"knowledge of letters", or music as (a kind of) "knowledge of sounds".
Something similar would seem to apply to the perceptible: vision may be
defined as the "perception of colours" (whereby "colour" is treated as a
species of "the perceptible").
To this third group also belong the relatives "the measurable and the
measure" (T6 fLETQ1TOV XLL TO ?TpQov). Every extension (or group of
things) can be measured (counted) by a conventionally adopted standard of
measurement (To V, To tEQOV), thus giving rise to a definite quantity.
Definite quantities may thus be legitimately considered as species of "the
measurable" (a relative), although they themselves belong to the category of
quantity (To noo6v, "how much", 5bl2). As Adolf Trendelenburg pointed
out (1846, 122), it should not be feared that "the measurable and the mea-
sure" might be confused with the first group of relatives: in the first group,
he says, the "standard of measure" (To [uErQOV) functions as a tacit means
of comparison of the two relata, whereas in the third group it represents one
of them.
269
I strongly suspect that the definitional priority of one of the relata in each
pair of relatives is a property common to all members of the third group in
Met. A 15, although Aristotle does not explicitly say so. Plato, in his Char-
mides (168b ff.), mentioned as examples of relatives - along with knowl-
edge and perception - will, love, fear, opinion and instinct. It is worth
noticing that all these relatives bear an internal relation to their respective
correlates.
After these considerations, some examples from the Categories, like "the
rudder and the ruddered" (7al2), "the wing and the winged" (7a4), will be
easily seen to fall under this third class, namely, as specifications of the
relatives "the part and the whole". In "the part and the whole", the relata
have an asymmetrical way of referring to each other: the part is defined by
reference to the whole (as fulfilling a particular function in it), but the
whole is not defined by reference to the part. Since relatives of the third
group may cease to be such once their determination has taken place, the
parts of substances (like hand, head, etc.) do not really belong to the cate-
gory of the relative.30
Again, the reason "position" is relational, is not, contrary to what an
initial survey of Categories 7 might suggest, the normal language usage, but
the fact that Aristotle presumably avails himself of this term as a relational
genus in order to define particular positions (cf. 6b 11 ff.); thus, "lying"
could be defined as "the position of an animal that rests horizontally"
(whereby "position" would be operating as a triadic relational predicate).3"
But the particular positions themselves (lying, standing, etc.) would likely
belong to the category he introduces as position (xFiOfat lb27). Similarly
with "state" and "condition": these terms are probably used as indefinite
genera to define particular states and conditions (Topics 121b36 ff., taken
together with 125a34-bl, suggests that knowledge has to be defined as "a
[such and such] condition of the rational part of the soul", or something
similar).
30 Nor are "wing" and "rudder" relatives, for similar reasons. Their unlucky inclusion in
the Categories as examples for relatives has been a constant source of confusion in
commentaries.
31 For a tacit recognition of triadic relations by Aristotle, see the examples "transfer",
"greater" and "gift" at Top. 125al6 ff.
270
v
I would like to conclude by reflecting on some metaphysical implications of

the results obtained in the preceding pages.
Since Aristotle allows for a category of "that which is said in relation to
something else" (tax rrQog ti), he may be considered as an adherent to the
thesis of the reality of relations. On the other hand, he describes relatives as
being those entities which are the "least real" of all (1088a29), because a
relational predicate may change without the entity itself changing. As we
have seen, this happens because Aristotle does not include the correlate in
the relational attribute, that is, because he considers the relational attribute
as indefinite. In taking as an initial guide for his categorial system linguistic
predicative practice, he notices that some attributes behave so that their
correct predication depends not only on the nature of the thing denoted by
the grammatical subject but also on the nature of other individuals sur-
rounding it.
Aristotle could hardly have overlooked the fact that substantial, qual-
itative and quantitative determinations are intrinsically connected with the
possibility of ascertaining the sameness, similarity or equality of things (as
well as their differences); he must have also noticed that the way in which
some things affect others constitutes an integral part of their natures (i.e. as
potentialities), and that such phenomena can only be apprehended by means
of relations. These are presumably some of the reasons for including in
Books F and I of the Metaphysics the most general pairs of relatives (such
as "the same", "the other", etc.) among the concepts which a projected
Science of Being qua Being would have to study.
But to infer from all this that Aristotle's system of the categories could be
developed from the starting point of mere relations (or, for that matter, from
contrary pairs of relational terms) would be to incur a serious mistake.
Aristotle makes clear that the predication of any relational attribute always
presupposes an underlying nature; he thus repudiates the idea that relations
be postulated as principles of Being (1088a20 ff.).
Relations fulfil another important role besides connecting the universe of
individual substances into a coherent whole, a x6o'pog. They help explain
how some entities can be defined by recurring to different, prior things,
without thereby losing their ontological autonomy. If definitions are to re-
veal the real structure of things, the relations involved in their formation
must also carry a reality of their own.
Both Plato and Aristotle avail themselves of relations in order to define
entities, employing thereby a natural procedure which is probably previous
to any established definitional methodology. They conceive some genera in
271
analogy to "normnal" relational terms (which are in themselves indetermi-
nate but can nevertheless be precisely predicated once their proper corre-
lates are specified).
However, the analogy between relational genera and ordinary relational
terms has a limit. When predicating a normal relational term of a subject,
the subject is essentially what it is, independently from the relation in which
it stands. This aspect of the analogy cannot be extended to relatives of the
third group in Met. A 15. For instance, the function of the hand in the living
organism is not independent from this organism's own nature, and thus the
hand cannot be considered as the thing it is, previously to its relation to the
body. If we separate as two different entities the relational genus from its
determining aspect, we shall sooner or later have to face the problem of
explaining how they can form a unity at all.32
I think it is the desire to avoid this difficulty which moves Aristotle to
distinguish in Met. A 15 the third group of "relatives" from the first two
groups. In contrast to conventional relations (first and second groups), in
which two independent things are always involved, the relation is, for rela-
tives of the third group, "internal" for one member of each pair (one of the
relatives is "prior" to the other in the sense of permitting its essential deter-
mination).
Relational terms may thus serve to draw attention to essential aspects of
the entities to which they refer. "Part" and "matter", for instance, are words
whose meaning betrays incompleteness. We usually say "part of some-
thing", "matter for something". By including terms as "part" and "matter"
in his metaphysical picture of the world, Aristotle seems to be reminding us
that there is no part or matter that can subsist by itself. At the same time, he
is implying that awareness of the relational character of these concepts is
basic for a proper understanding of the broader context in which they are
used.
It should be borne in mind, however, that the indefiniteness of relational
genera is not always evident. This may be due to the fact that: 1) linguistic
expressions are sometimes ambiguous, 2) our list of essential relations va-
ries according to the ontology we sustain, 3) genera refer extensionally to
the individuals falling under them, which in themselves are always determi-
nate, so that the indefinite character of the genus might easily pass unno-
32 In Met. Z 12 Aristotle shows that the genus behaves to each of its differentiae, in
definition, as matter does to form, in the concrete individual. Since "matter" is for
Aristotle a nQ6g Ta (Phys. 149b9 f.), this seems to imply that all genera reveal them-
selves on closer examination as being relational in character, i.e. they remain indefinite
until they are determined by their differentiae.
272
ticed. Aristotle seems to have been conscious of these facts, for he occa-
sionally insists on the relational character of some terms and their lack of
determinacy. Trendelenburg observed (1846, 127) that the concept of
oxFQoiL bears an essential relation to its respective substrate, and that
"matter" is explicitly described at Physics 149b9 f. as being a rQog tL (no
doubt because, as we just said, every kind of matter has to be determined as
matter for/of a [specific] form). But he erroneously included "form" among
the "relatives" (loc. cit.), since he failed to perceive the structural asymm-
etry peculiar to the relation form/matter (and, as suggested, to all pairs that
belong to the third group of relatives in Met. A 15). The fact that we do use
the expression "the form of something", although "form" is not a relative,
proves how misleading language can be.33
Let us briefly consider the well known example of "snubness" in Met. Z
5. Here "snubness" is agreed to mean "hollowness of the nose". Now, in my
opinion, "hollowness" functions here as a relational, indefinite predicate,
and "of the nose" as its specification. The redundancy Aristotle criticizes in
this chapter is not due to the fact that "snubness" is defined as "hollowness
of the nose". There is nothing wrong with this. A redundancy only results if,
in defining an individual like the (snub) nose, one includes, so to say, both
the substance (nose) and its accident (snub), because, since in snubness the
notion of nose is already presupposed, "snub nose" would be equivalent to
"nose that is a hollow nose". I believe that for Aristotle all accidents imply,
sooner or later, a determnination analogous to the one involved when deter-
mining relational predicates, i.e. all accidents have to be determined as
modifications of an underlying subject, and thus presuppose it.34
But again, why should anyone wish to define the substance together with
its accidents? Perhaps because, according to Aristotle, it is the task of a
philosopher not only to define each entity by itself, but also to explain how
entities relate to each other in the actual world. Now, if one were to grant
substances and accidents the same ontological status by reason of the fact
that they can both be defined - as they indeed can -, then it would be
impossible to explain their coexistence, because a "redundancy" would in-
evitably ensue (in the way shown in Met. Z 5). To treat both a quality and a
substance as entities in the same sense, confronts us with the problem of
making intelligible how "two different things" can become "one" (sub-
stances do form, after all, a unity with their accidents). Aristotle's solution
is to reserve true definability for substance, and a secondary type of defin-
33 The traditional view that definition D2 of relatives is more ontological than linguistic
(vide supra) is, therefore, justified.
34 See Met. Z 1, 1028a35 f.
273
ability for accidents as well as for the compounds of substance and acci-
dents.
One can think of other possible implications of Aristotle's conception of
relational attributes for his metaphysical system. It would be worth in-
vestigating if his notion of "focal meaning" at the beginning of Metaphysics
F 2, the so called tQo6; tv relation of heterogeneous things, can be partially
explained in the light of his peculiar theory of relations. It seems obvious
that the focal item is definitionally prior to the defined entities. Now, since
apparently unconnected things have to be determined by reference to one
and the same notion, perhaps it is legitimate to infer that these things do
have an affinity after all (in the same way, for instance, in which knowledge
has an affinity, i.e. form, with the reality it reflects). This in turn could serve
as a clue as to where to look for the much needed "unity" of the Science of
Being qua Being.35
Universidad Simon Bolivar, Caracas, Venezuela.
References
Ackrill, J.L., 1963, Aristotle's "Categories" and "De Interpretatione", Oxford, Clar-
endon Press.
Barnes, Jonathan, 1984, The Complete Works of Aristotle, The Revised Translation, Ed.
by J. Barnes, 2 vols., Princeton, Princeton Univ. Press.
Cornford, F.M., 1935, Plato's Theory of Knowledge, London, Routledge.
Erler, M., 1992, "Relation. 1. Antike", in: Historisches Worterbuch der Philosophie, vol.
8, Darmstadt, Wissenschaftliche Buchgesellschaft.
Kahn, Charles, 1978, "Questions and Categories", in: H. Hiz (ed.), Questions, Dor-
drecht, Reidel Publishing Co., pp. 227-278.
Minio-Paluello, L., 1949, Aristotelis Categoriae et Liber de Interpretatione, Oxford,
Clarendon Press.
Ross, David, 1924, Aristotle's Metaphysics, 2 vols. Oxford, Clarendon Press.
Scheibe, Erhard, 1967, "Uber Relativbegriffe in der Philosophie Platons", Phronesis vol.
XII, pp. 28-49.
Trendelenburg, Adolf, 1846, Geschichte der Kategorienlehre, Berlin, Bethge Verlag.
35 I prefer to use "affinity" instead of "common feature" or "common property", because

I have reasons to believe that fully heterogeneous entities - like those involved in a JTQog
tv relation - cannot have, from Aristotle's viewpoint, common features or common
properties at all. As one of my students, Alexander Medina, pointed out to me, this
would imply that the Science of Being qua Being cannot be understood as the science
that studies the properties common to all entities, since there are no such properties.
274

Relational Attributes in Aristotle

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Relational Attributes in Aristotle

Caricato da

Copyright:

Formati disponibili

Relational Attributes in Aristotle

Author(s): Fabio Morales

Aristotle's theory of relations involves serious difficulties of interpreta-

The most extensive treatment of relational attributes by Aristotle is to be

Phronesis 1994. Vol. XXXIX13 (Accepted January 1994)

For convenience and clarity of exposition, I will begin my discussion with

"Of' or "than" represent in this translation the Greek genitive, which is

2a) a subject of which the relational predicate is predicated,8 and

Towards the end of Categories 7 Aristotle mentions the possibility that

The interpretation of definition D2, as well as the difference between it and

'" See, for example, Aristotle, Met. F 6, 101 I bl7.

"1 "[...] a severed finger is a finger only in name", Met. 1035b23-25.

21 Translation by W.A. Pickard-Cambridge (Barnes, 1984, I), my italics.

In order to make the mentioned distinction between two fundamentally

2) Relatives said to be "according to a capacity" (xa'rit 66v%ttv). Some

I would like to conclude by reflecting on some metaphysical implications of

Universidad Simon Bolivar, Caracas, Venezuela.

35 I prefer to use "affinity" instead of "common feature" or "common property", because

Potrebbero piacerti anche