The following text has been automatically reproduced by an Optical Character Recognition (OCR) algorithm. It may not have been checked over by human eyes. For matters of precision please consult the original pdf.

Abstraction: A Realist Interpretation

Abstrac:tion: A Realist
Andrew Saver

The relations between the theoretical and the
empirical, the abstract and the concrete, have always
been problematic in marxism. Marx’s disdain for
knowledge based upon mere appearances has meant that
few marxists have accepted the empiricist doctrine of
the theory-neutrality of observation. But while, in
a negative way, there is a consensus about the rejection of this doctrine among marxists, and while we
often quite readily talk of ‘essence and appearance’

and ‘l’lnderlying’ structures and causes, there is
little agreement about an alternative view of the
status of marxist concepts and of the relations
between the theoretical and the empirical. The
radical undermining of empiricist views on this
relation in the philosophy of science has been similarly unsettling, producing shifts towards idealism,
particularly in the form of conventionalism. The
abandonment of the dangerous innocence of certainty
in knowledge based on experience has given way to
possibly more dangerous views in which knowledge is
believed not to be subject to any extra-discursive

This crisis at the philosophical level has surely
made its impact on substantive marxist research. A
major characteristic of recent marxist study has been
a withdrawal from empirical research and a turning
inwards towards a continual reconstitution of abstract
theoretical concepts (even where new objects of study
– such as the state – are concerned), or else a kind
of ‘pseudo-concrete’ analysis where the specificities
of the concrete are reduced to an abstract category.

It is not too much to say that for some the recognition of the impossibility of theory-neutral observation has induced a fear that any empirical research
would inevitably be tainted by empiricism.

An early opponent of this anti-empirical or
‘pseudo-concrete’ tendency was Sartre:

‘There is no longer any question of studying facts
within the general perspective of Marxism so as to
enrich our understanding and to clarify action.

Analysis consists solely in getting rid of detail
and forcing the significance of events’ [1]
and, more strongly:

‘Marxism possesses theoretical bases, it embraces
all human activity; but it no longer knows
anything. Its concepts are dictates: its goal
is no longer to increase what it knows but to be
itself constituted a priori as an absolute
knowledge’ [2].


A strikingly similar kind of criticism is made in
many of Raymond Williams’ writings. For example, in
Marxism and Literature, he attacks the kind of marxism in which:

‘ … the analytic categories, as so often in
idealist thought, have, almost unnoticed, become
substantive descriptions, which then take habitual
priority over the whole social process to which,
as analytic categories, they are attempting to
speak.’ [3]
And again, in less sober style but with similar
intention, E.P. Thompson has polemicisedOagainst a
condition which he aptly terms ‘intellectual agoraphobia’ [4] epitomised by those for whom the concept
‘mode of production’

‘ … has become like a base camp in the Arctic of
Theory which the explorers may not depart from for
more than a hundred yards for fear of being lost
in an ideological blizzard.’ [5]
This kind of reductionism is common to many areas
of marxist analysis, whether economic, political or
cultural. It is politically damaging because the
failure to grasp the specificities of the concrete
inevitably weakens attempts to inform practice.

Practice always takes place in the muddy waters of
the concrete: it cannot be usefully informed by a
theory which does no more than reduce the concrete
to the abstract.

But all this is no more than a statement of the
problem. To solve it, it is at least necessary to
clarify concepts such as ‘theoretical’, ’empirical’,
‘abstract’, and ‘concrete’. This paper attempts this
by drawing upon arguments from the realist theory of
science, especially as it has been recently developed
by Bhaskar and Harre [6]. In so doing, I shall try
to shift debate about these concepts outside the
crippling polarity of empiricism and rationalism
which characterises the present crisis of epistemology.

Theory and Observation: preliminary points
It is now widely recognised that observation is not
theory-neutral but theory-laden, and that theory does
not merely ‘order facts’ but makes claims about the
nature of its object. So, in evaluating observations
we are also assessing particular theoretical concepts
and existential claims. A common response to this
shattering of innocent beliefs in the certainty and

neutrality of observation has been the development of
idealist (especially conventionalist and rationalist)
philosophies which assume that if observation is
theory-laden, it must necessarily be theorydetermined, such that it is no longer possible to
speak of criteria of ‘truth’ or ‘objectivity’ which
are not entirely internal to ‘theoretical discourse’.

However, this is a non-sequitur for at least two
reasons. First, theory-laden observation need not
be theory-determined. Even the arch-conventionalist
Feyerabend (1970) acknowledges that ‘it is possible
to refute a theory by an experience that is entirely
interpreted within its own terms’ [7]. If I ask how
many leaves there are on a tree, my empirical observation will be controlled by concepts regarding the
nature of trees, leaves and the operation of counting,
but to give an answer I’d still have to go and look!

In arguing that there are no extra-discursive criteria
of truth, recent idealists such as Hindess and Hirst
echo Wittgenstein’s identification of the limits of
our world with the limits of language, and share the
confusion of questions of What exists? with VVhat can
be known to exist? The truism that extra-discursive
controls on knowledge can only be referred to in
discourse does not mean that what is referred to is
purely internal to discourse [8]. Secondly, and
more simply, it does not follow from the fact that
all knowledge is fallible, that it is all equally

While recognition of the theory-laden nature of
observation suggests that any rigid distinction
between description and explanation should be
abandoned, we presumably would wish to retain a
distinction between theoretical research (or critique
or reflection) and empirical research. Certainly
empirical research can never be a-theoretical, but it
would seem to be a different activity from theoretical

Abstract and Concrete
To try to provide a sound basis for the distinction:

theoretical/empirical it is necessary to consider a
related, but not identical, distinction that is
fundamental to marxist method: that between the
abstract and the concrete.

Marx’s own definition of the concrete from the
1857 Introduction has been trotted out in scores of
recent marxist writings but is worth examining to see
how it differs from the more familiar concept of the
’empirical’ .

‘The concrete concept is concrete because it is a
synthesis of many definitions, thus representing
the unity of diverse aspects.’ [9]
By ‘concrete’ we mean something real, but not something which is reducible to the empirical: we mean
far more than just ‘factual’. The concrete object is
concrete not simply because it exists, but because it
is a combination of many diverse forces or processes.

In contrast, an abstract concept represents a onesided or partial aspect of an object. For example,
if we conceptualise an object such as a factory
simply in terms of its outward appearance, the concept will be abstract in the sense of one-sided even
though it refers to something which can be empirically
observed. To make this a concrete concept we would
have to specify all the relationships in which the
factory is involved: with its workforce; its suppliers
and buyers; its creditors and competitors, etc. These
diverse determinations are not simply listed and
‘added up’, hut arc s)~thesised; that is, their
combination qual i t;Jt i “”e 1y modifies each constituent
element. However, in order to understand this combination, we normally have to isolate each element in
thought first, even though they do not and sometimes
could not exist in isolation in reality. It’s
important to note that whether the concrete is

observable (and hence an empirical object for us) is
contingent (i.e. neither necessary nor impossible).

The concepts ‘concrete’ and ’empirical’ are not

What is then awkward is that Marx also sometimes
uses the term ‘abstraction’ pejoratively. Again, in
the 1857 Introduction, he discusses various ways of
studying the political economy of a country [10].

The possibility of beginning with the population is
dismissed as an abstraction unless it is broken down
into its constituent classes, for in concealing
these, it would be a ‘chaotic conception’. So evidently there are good (rational) and bad (‘chaotic’)
abstractions. It would take quite a long discussion
of marxist theory to demonstrate why it is essential
to deal with classes rather than population or, for
that matter, any other aspect of the population.

Without such a defence, Marx’s criticis~ is liable
to appear to the non-marxist as simply a dogmatic
assertion. What is required here is surely a general
epistemological distinction for discerning misleading
abstraction from enlightening or rational abstraction:

the abstract-concrete distinction is not enough on
its own. Moreover, as we shall see, it doesn’t help
us distinguish between what can be known from
theoretical analysis and what must be learned from
(theoretically-informed) empirical study. To try to
solve these problems, I shall draw upon some recent
work in the realist philosophy of science.

One of the most direct challenges realism makes is
on the question of Hume’s problems of causation and
induction. Starting from an ontology of discretelydistinct, atomistic events and objects, Hume
insisted that there could be no necessary connexions
between these. We might observe regularities in
patterns and sequences of events, but any attributions of causal connexion could only by of psychological origin, for knowledge that C has always been
followed by E in past experience does not logically
guarantee that it will always do so.’ Even if we
could establish that constant conjunctions were universal, they would still be contingent. Causation
is therefore equated with regular succession, and so
cannot be distinguished from correlation or accidental

This counterintuitive, but logically sound argument
concerning the problem of induction has come to be
known as the ‘scandal of philosophy’, for it would
seem that we are perfectly capable of distinguishing
between the causal processes that make the hands of a
clock move, and the accidental relationships that
might arise between the Swiss bank rate and the
Australian divorce rate. If we take the ‘scandal’

seriously, then (pace Popper) neither verification
nor falsification can be of any use, for without any
necessity in nature, what is confirmed today may be
falsified tomorrow – and vice versa! [11]
Realists have argued that, although it is logically possible that the world itself may suddenly
change completely (the ‘big’ problem of induction),
this does not mean that everything in our present
world is contingently related [12]. If all objects
or events are independent, then their pattern or
succession is certainly accidental, but precisely
because some changes are changes in things, not all
changes are independent or accidental [13]. In other
words, an atomistic ontology makes it impossible to
distinguish between the concepts of a change in the
nature of a thing and successive replacements of the
thing, with the consequence that regularities have to
be treated as accidental persistences of events for
which there is no rational explanation [14].

Realists dispense with the Humean metaphysical
predilection for atomism, and causation is understood
instead as the necessary ways-of-acting of an object
which exist in virtue of its nature. That is, causation is not conceptualised in terms of a relationship

between separate events ‘C’ and ‘E’, but in terms of
the changes in each of ‘C’ and ‘E’. Gunpowder has
the ‘causal power’ to explode in virtue of its unstable chemical structure. Copper can conduct
electricity because of the presence of free ions in
its chemical structure. Whether either of these
causal powers are ever ‘realised’ or ‘activated’

depends upon contingently related conditions, such as
the presence of oxygen, low humidity and a spark in
the first case, and an electric current in the second.

Because the conditions are independent of the causal
powers, the succession of events cannot be known just
on the basis of knowledge of the causal powers. So
it is contingent that gunpowder ever explodes, but in
certain conditions it will do so necessarily.

Scientific ‘laws’ are therefore not understood as
well-corroborated, universal empirical regularities
in patterns of events, but as statements about

‘The citation of a law presupposes a claim about
the activity of some mechanism but not about the
conditions under which the mechanism operates and
hence not about the results of ,its activity i.e.

the actual outcome on any particular occasion. J
The essential characteristic of law-likeness is not
universality but necessity. This necessity has
not·hing to do with the logical necessity which may
hold in the relationships between statements, for
what happens in the natural world has nothing to do
with the statements we use to describe it. Rather,
it is natural necessity. By this we mean that a
particular substance or object could not be what it
is unless it had that particular power, that way-ofact~ng.

If a substance cannot conduct electricity,
i t certainly cannot be copper. It is logically
possible that the world – including copper- may
suddenly change into something different, but while
the substance is still copper it must have these
causal powers and a specific nature. That this is
not simply a matter of tautology will be explained

The realist account of scientific laws is compat-·
ible with the marxian notion of laws as tendencies.

The law of value does not refer to an empirical
regularity, nor a generalisation, nor a trend, but a
mechanism which operates in virtue of the competitive
nature of capitalist commodity production. The
effects produced by it at the empirical level depend
upon contingently related conditions, including those
produced by other mechanisms which are sometimes
called ‘counteracting tendencies’. In the case of
many of the tendencies of marxist theory, surplus
empirIcal information has to be gained in order to
know how the mechanism itself is operating. For
example, it is necessary that, given capitalist
relations of production, the law of value will produce a lowering of the value of commodities over
time. But the rate of this lowering for different
commodities is affected by use-value considerations
– in particular the kind of use-values demanded and
the kind of technologies available to produce them and considerations of class struggle in terms of
value as a social relation.

On this realist account, there is no presumption
that real relations are structured like conceptual
relations and so epistemological legislations founded
upon logical relationships are considered to be unhelpful. Nevertheless, it is argued that it is
possible that relations between concepts can be made
to map real ones [16]. Although the real object is
quite separate from the thought object, this does not
rule out the possibility that some sort of ‘correspondence’, or relation of ‘practical adequacy’, can
be achieved between the two [17]. In other words
realism neither assumes an epistemologically
privileged observation language which guarantees


correspondence to the real object nor falsely assumes
that the lack of a theory-neutral observation language
means that observation is completely theory-determined
such that there can be no correspondence whatsoever.

Take the example used by Harr~ and Madden (1975)
of the definition of the term ‘father’. It is true
by definition that a father (in the biological sense)
is a man who has or has had a child. However, the
conceptual necessity here is used to denote an
empirically-discovered natural necessity in the relationships between males and procreation. Apparently,
certain aborigine peoples are not aware that the male
has any role in procreation and so do not have any
equivalent in their language for the word ‘father’.

When we discover such natural necessities we frequently make what were previously understood as contingently related elements part of the definition of objects;
indeed, one might say that progress in science, in
terms of reduction of the burden of facts, depends on
this [18]. That a father is a man who has or has had
children is not just a tautology, for if it were,
science could develop simply by inventing tautologies
freely at will. But it is always an empirical question whether any real object is like our definitions.

In this way, natural necessities can be ‘taken up’

into the language in the form of conceptual

‘Should the relation between the nature of an
entity and its powers be naturally necessary, we
hold this to be an a posteriori truth about the
entity, and so it must be the case that in that
world such an entity is capable of an alternative,
earlier and more naive description, under which
its nature thus described is merely contingently
related to those of its powers and liabilities
which are later discovered to be necessary consequences of its real nature.’ [19]

Not all natural necessities that we discover are
‘taken up’ into the language in the form of conceptual
necessities, for some can be described by contingently
related statements [20]. It is necessary that we eat
and satisfy certain physical requirements if we are
to survive, but this natural necessity has not been
‘taken up’ into the definition of human beings,
probably for the good reason that it would not differentiate us from animals. It might also seem possible
that a capitalist could stop purchasing labour-power,
stop accumulating capital and therefore break the
necessary relationship between these actions and being
a capitalist, but in acting this way slhe would be
becoming a non-capitalist. In marxism, these necessary relationships are ‘taken up’ into the definition
of capital [21], but there are other claims about
natural necessities which have a simpler description,
but which are implicit in the theory nonetheless.

As Harre and Hadden suggest, relationships which
were once considered to be independent may later come
to be recognised as necessary. Yet as some recent
trends in marxist theory have shown, progress has, in
some instances, consisted in showing that certain
(sets of) relationships which were formerly seen as
necessarily linked are now known to be only contingently related or capable of a wider range of forms
of combination than was previously realised. This is
true of historicist, stage-theory notions of

Also, as Banaji shows, the concept of modes of
production can be inadequate both as an abstract or
a concrete concept because it is now realised that

modes of production are not nearly as limited in terms
of possible forms of interlocking combinations of
relations and forces of production as was originally
thought [22]. It seems, therefore, that the concept
‘modes of production’ can be given a less crucial
summarising role. It is more important to establish
the actual combinations of forces and relations of
production that exist and work out how they cohere
and function. Trying to force aberrant facts into
simple categorisations of feudal or capitalist or even
into ‘articulations’ of several modes of production
(each of whose form can be known in advance from
theory) by arguing that the facts must have been
theorised incorrectly is neither useful nor necessary.

Banaji shows that restricted, idealised views of
modes of production have inhibited the development of
marxist theory of the transition between feudalism
and capitalism and Third World social formations.

The consequences of relegating the concept to a lesser
theoretical role need not be damaging (surprising
though it may seem) for the essential notions of a
relatively enduring interlocking of relations between
people, and between people and nature can be retained
using lower-order concepts or at least less restrictive formulations of ‘mode of production’ than is
found in much marxist writing.

We can now clarify the relationship between the
abstract and the concrete, and also the distinction
between good and bad abstraction. Good or ‘rational’

abstractions should isolate necessary relationships.

The concrete, as a unity of diverse determinations,
is a combination of several necessary relationships,
but the form of the combination is contingent, and
therefore only determinable through empirical
research. As such~ its form cannot be assumed to
have already been ‘taken up’ into the theoretical
framework in the same way that the nature of the
abstract can.

A bad abstraction or ‘chaotic conception’ is one
which is based upon a non-necessary relationship, or
which divides the indivisible by failing to rec~gnise
a necessary relationship. The same point can be made
in a different way by using the distinction between
external and internal relations. The relation between
a person and a lump of earth is external and contingent in the sense that each object can exist without
the other. On the other hand the relations between
landlord and tenant, master and slave are internal
and necessary in that what each part of the relation
is depends upon its relation to the other. Sometimes
internal relations may be asymmetric as in the case
of state and council housing, money and banking system
in which the former object in each pair can exist
without the latter, but not vice versa [23]. A
rational abstraction – unlike a chaotic conception takes due account of structures of internal and
external relations.

Theories make their strongest claims at the
abstract level, about necessary and internal relations, about causal powers which exist in virtue of
the nature of particular things. They quite properly
remain more agnostic towards the form of external
relations. Physics quite rightly makes a strong
claim about copper’s power to conduct electricity,

but does not commit itself on whether any particular
piece of copper ever will be in a position to do so.

And similarly with marxism; given that capital
cannot exist as such without wage-labour, we should
not develop abstractions which treat them as independent. If General Motors could function in its present
form with serf labour, the theory really would be in
trouble, but it quite properly does not commit
itself on the contingent matter of whether that
labour is American, British or Turkish. We may make
theoretical claims about the former and agree that
confirmations or falsifications are epistemically
significant, but the testing of empirical claims made
about contingently related processes need not affect
our confidence in the theoretical claims. It may be
important to establish what proportion of General
Motors’ labour force is American, but if we get it
wrong, this is unlikely to warrant a challenge to
basic theory.

This is not of course to say that concrete objects
are unimportant – far from it; but what theory provides us with is an understanding of the concrete by
means of abstract concepts denoting its determinations. In this context, the primary position of the
concept of ‘commodity’ in Capital is rightly noted in
making the point that, although abstract concepts
have to be used to explain the concrete, we have to
start with what we have to explain. But the major
theoretical issues are not about a simple category of
the commodity, as it might be instantiated in, say,
a car, but about the abstraction of use-value and
exchange-value as its essential determinations.

In Bhaskar’s terms rational abstractions concern
the level of the ‘real’ – causal powers or generative
mechanisms; concrete concepts concern the level of
the ‘actual’ – the effects, operation and activation
of mechanisms, it then being contingent whether these
are possible empirical objects for us [24].

Figure 1 sums up the hierarchy of types of concepts in marxism ranging from the most basic abstract
concepts which refer to transhistorical necessities,
through historically-specific abstract concepts,
through the ‘tendencies’ which are the equivalents
of ‘mechanisms’ in realist philosophy of science, to
the more concrete ‘level’. As we have seen, because
of the historical nature of society, which historically-specific abstractions must be used depends upon
the kind of basic necessary relationships which
obtain at any point in time. In natural science,
natural necessities are empirically discovered too,
but in general, they do not change. And this is why
marxism (indeed, any social theory) cannot take its
more basic concepts for granted to the extent that
natural sciences can: the concepts must change with
the reality they depict, or of which they are

Although we can say that certain necessary relationships in capitalism have been ‘taken up’ into
marxist theory in such a way that we can ‘know in
advance’ that wherever there is capital, there must
also be value-producing wage-labour, it must be
stressed that this knowledge is ultimately grounded
a posteriori. In like manner, given the existence of
a child, we can ‘know in advance’ of the existence of
a father, but even this knowledge is, as we have seen,
an a posteriori discovery of a necessary connexion.

So even the most basic theoretical claims at the top
of the diagram are in principle revisable; they are
not to be taken on faith. Necessary relationships
may exist in reality but it is contingent whether we
know them [25].

In moving down the diagram towards the concrete,
knowledge of contingently related phenomena must be
combined with knowledge of abstract necessities.

These contingent relations are affected by: (a) class
struggle – which can also change the structures in
virtue of which mechanisms or tendencies operate;


(e.g. concepts of people and nature)

(e.g. teleology of labour, social
relations of production)






(e.g. capital-wage labour)



xl’ x 2 ‘·····,xk (e.g. law of value)


(including other tendencies)

(‘unity of diverse aspects’) to form
CONCRETE ~ z2’· … ·, zk


(within – in Bhaskar’s terms – ‘open systems’)
* The theorisation of these, and their explanation by means of abstraction,is often not the sole prerogative of marxism.

(b) theory itself, as in praxis; and (c) future
knowledge, which, for the reasons given by Popper,
is unknowable now. Some of these conditions may be
satisfactorily theorised outside marxism, others may
need re-theorising. And of course, as the example
of the critical insights on marxism generated by
feminism show, the direction of conceptual change
need not be one-way. So, for example, although we
can say from basic theoretical propositions about
capitalism that the law of value forces a continual
restructuring of capital upon firms, we cannot know
in advance what form that will take because it
depends, among other things, upon the nature of technology, which in turn depends on the growth of
knowledge; we simply have to go and find out through
theoretically-informed empirical research.

Marx’s own position on this movement down the
diagram from the abstract to the concrete is ambiguous in terms of whether he overlooked this essential
element of contingency. What Marx considered to be
‘ … obviously the correct scientific method …

[led] … from abstract definitions by way of
reasoning to the reproduction of the concrete
si tuation.’ [26]
‘By way of reasoning … ‘ suggests a movement which
is purely internal to thought, which cannot or need
not ask empirical questions [27]. However, elsewhere
in the Introduction, Marx says that the ”’concrete in
thought” is a product of the working-up of observation and conception into concepts’ [28]. The latter
interpretation is the only one which can make sense
of his concrete historical studies. So, movement
between the levels of the diagram does not generally


involve moves of deductive logic. To move from transhistorical theoretical claims (e.g. ‘All production is
carried out under social relations’) to historicallyspecific claims (‘capitalist production requires a
propertyless class of workers’), we have to add
historical information which is not implicit in the
premises of the transhistorical claims [29].

There is also a more general reason why these
things cannot be known in advance, and again it
depends upon a distinction between necessary and
contingent relations between things.

‘It is because things cannot be reduced to the
conditions of their formation that events are not
determined before they are caused to happen. This
fact accounts for both the temporal asymmetry of
causes and effects and the irreversibility of
causal processes in time.’ [30]
So, while everything is ‘determined’, things are not
pre-determined except where, as in experiments, conditions are controlled. The chemical structure of
gunpowder ‘determines’ its explosive causal powers,
but whether it ever does explode is not thereby predetermined.

Although the ‘knowledge’ represented in the
diagram centres on marxism, it should not be seen as
self-contained but as extending horizontally and
vertically into knowledges with different domains
such as the natural sciences and psychology. These
‘discourses’ are neither reducible to a single discourse nor are they discrete. Often discourses
which compete in the same domain share an agreement
or indifference towards certain concepts which they
both use. Non-marxists may accept some of the most

basic claims of historical materialism e.g. about
transhistorical necessities or about some limited
aspects of very specific concrete concepts, but the
‘penumbra of meaning’ of these concepts will vary
according to the other elements of their discourse.

Marxists, qua marxists, are unlikely to question
the technical knowledge of an engineer although they
may have different interpretations of the social
context of engineering. In neither case need there
be incommensurability between the two discourses, and
in the first case there may be mutual agreement and/
or indifference. In both cases marxists may have to
draw upon this non-marxist knowledge, to understand
their own concerns (e.g. restructuring of capital)
when moving from the abstract to the concrete.

The knowledge represented in the diagram also
reflects a ‘stratification’ of the real [31]. What
marxism may take as ‘a given’ (e.g. human anatomy)
may be the prime object of study for another subject
working upon a different stratum. The existence of
this stratification need not mean that every event in
society can only be explained through a regress which
goes back through these strata to some first cause,
because each stratum is, despite being constituted
through processes at another stratum, irreducible: it
has emergent powers. Just as water has powers irreducible to those of hydrogen and oxygen; just as
human beings as organisms have powers irreducible to
the chemical processes which constitute them, so
certain combinations of material and social relations
produce social structures which have emergent powers
[32]. And it’s in virtue of these emergent powers
that ‘higher stratum objects’ intentionally or unintentionally react back upon lower strata, not by
‘breaking’ natural necessities, but by exploiting
contingency at the lower levels.

(Although it is far beyond the scope of this paper
and its author to outline an aetiology of society
which would substantively specify this stratification,
a word of caution is needed to guard against any unexamined over-hasty reinterpretation of (Althusserian)
‘levels’ of the ‘economic’ ‘political’ and ‘ideological’ as distinct strata, for they may possibly be
more accurately seen as different parts of the same
stratum. )
It should also be noted that there is no necessary
correspondence between the abstractness (or ‘onesidedness’) of a concept or the ‘height’ of the
stratum to which it refers, and its social significance. It is only because we usually forget the
abstract nature of many commonplaces that we tend to
associate abstractness with ‘theoretical significance’. Adorno provides a convenient illustration:

‘The category “societies with a division of labour”
is of a higher and more general order than the
category “capitalist society”; but it is a less,
not a more essential one, with less to say about
the lives of human beings and what threatens them
– without this implying that a lower order category such as “urbanism” has more to say on the
subject. The degree of abstraction of sociological
categories varies neither directly nor inversely
with their contribution to the understanding of
society.’ [33]
On this view of theory, the conceptualisation of
necessary relationships is absolutely critical. The
identification of mechanisms depends upon careful
description of the objects and relationships in virtue
of which they act. In contrast, description of
entities is treated as an unimportant preliminary to
theorising in empiricism. ‘Facts’ are assumed to be
capable of simple, atomistic description, and
theoretical issues are seen as problems of ordering
these facts. In this way, many necessary relationships are overlooked or distorted; usually by tearing
obj ects from the context upon which they are dependent
and ignoring their historically-specific character.

One of the most striking things about Marx’s work
is the thoroughness of this basic description.

Exchange-value is examined in terms of what it ‘presupposes’ – private property, division of labour,
production of commodities etc. The passage in the
1857 Introduction documenting the ways in which
production, distribution, exchange and consumption
interpenetrate and presuppose one another is a particularly good example; in noting that distribution
was first and foremost a distribution of the means
of production, and hence a relation within production, Marx demonstrated the existence of an internal
relation [34].

From the standpoint of modern bourgeois social
science, this kind of analysis has an unfamiliar
aspect. For example, consider the following:

‘All production is appropriation of nature on the
part of an individual within and through a specific form of society. But it is altogether
ridiculous to leap from that to a specific form
of property e.g. private property.’ [35]
‘ … that there can be no production and hence no
society where some form of property does not exist
is a tautology …. ‘ [36]
‘The obvious, trite notion: in production the
members appropriate (create, shape) the products
of nature in accord with human needs: … ‘ [37]
Marx clearly regarded such statements as an essential
foundation, but also as unexceptional, ‘obvious’,
‘trite’. And yet so much of liberal social science
appears not to know of these ‘trite’ notions. Entire
social theories have been constructed in wAich
society is ‘organised’ but somehow not dependent on
the appropriation of nature for its existence; and
the treatment of production and distribution as
simply externally related is still common.

On first encounter, much of this repetitive, painstaking – even ponderous – description of basic
entities and relations does seem ‘t~ite’ as Marx
admitted: do we really need to be told that exchangevalue presupposes a division of labour; or that
language cannot exist for an individual? But it is
these foundations that provide a means of distinguish·
ing rational abstractions from the chaotic conceptions which characterise ‘sciences’ which adopt a
casual attitude towards initial conceptualisation, or
at worst, as in much of neo-classical economics,
reduce it to a matter of defining mathematical notation-”’K” is capital and capital is “K”, and let’s
get on with the model’.

Yet if we refer back to the quotations from the
Grundrisse and the passages in which they occur, an
ambigui ty in Harx’ s discussion can be seen. By
referring to the relationships as ‘tautologies’ in
which ‘categories’ ‘presuppose’ one another, it
appears that they are conceptual necessities and
nothing more. The theory has the appearance of what
Marx himself called an a priori construction. Nevertheless, consideration of instances of these conceptual connexions shows, as we have seen, that they
are based on real, necessary connexions [38]. And
indeed, many theories appear to be largely a priori
constructions. This is not unusual, and it need not
be a cause for concern unl-ess, like Humeans, we
render natural necessity unintelligible by adopting
an atomistic ontology and hence make it impossible to
recognise that some conceptual necessities have a
real basis [39]. The important question is How, if
at all, are they grounded in necessary relationships?;
How are the latter ‘taken up’ into the theoretical
concepts? [40]. Provided that those relationships
between objects which are genuinely independent are
not treated as necessary connexions, so that empirical questions are prejudged in an a priori manner,
the generally a priori character of a theory need not
be a problem.


Some of the most basic elements of Marx’s
critique of political economy concern his corrections
of confusions about the real which arise from the
misleading logical structure of discourse. Interpreted in cornrnonsense fashion the concepts ‘production’ and ‘distribution’ do not mutually presuppose
one another: their relation is not analytic but
synthetic. It is only when each of these concepts
is ‘unpacked’ and their objects examined in their
material contexts that it becomes clear that they
denote necessarily or internally related objects.

We can now go back and give further clarification
of the proper meaning of ’empirical’ and’ theoretical’.

The .Empirical
Implicit in the above crltlque of empiricist ontology
and the discussion of laws of tendency was an attack
upon the concept of an ’empirical world’. Interpreting ’empirical’ here as ‘that which is observable’,
the concept of ’empirical world’ arises from an
illegitimate reduction of an ontological question to
an (empiricist) epistemological one. Now it would be
extraordinary if ‘the real’ just happened to be
exactly coextensive with the limits of our sensory
powers. This solipsistic exclusion of a nonempirical real world also generates a whole range of
problems, one of the most obvious of which is that
of understanding how we ever come to discover anything new. Moreover, as we have seen, it also
secretes a notion of an identity of thought-object
and real-object, and therefore implies a completed
science grounded in certain or absolute empirical
knowledge. However, if we accept that observation
is theory-laden, such that no clear distinction
between what can be observed and what can be inferred
on the basis of observation can be sustained, then we
must acknowledge that the boundaries of ‘the empirical’ are both fuzzy and changeable. Wllat is empirical depends upon our knowledge and sensory powers:

what is concrete (excluding conceptual objects) does

Marxism’s distinction of essence and appearance
and its rejection of empiricist epistemology are incompatible with the ’empirical world’. Where marxists have attempted to reject this epistemology while
retaining its flat, unstratified ontology, the result
has been idealist contortions where the essential and
the abstract are denied any reference to real objects
and are reduced to heuristic devices for understanding the empirical. In the philosophy of science, the
recent development of conventionalist critiques of
positivism have been based on this same incompatible
combination [41].

If we do not accord real status to mechanisms and
instead treat laws as statements about universal
empirical regularities we run into what Bhaskar terms
a dilemma of ‘actualism’ [42]. Faced with the conspicuous rarity of spontaneously-occurring, precise,
universal empirical regularities, we can either:

(a) conclude that any contenders for the status of
‘law’ are thereby refuted, or
(b) conclude that laws apply only to ideal conditions
equivalent to those of scientific experiments,
and nowhere else.


It is only if laws are understood as referring to
mechanisms, and not empirical events, that the manifestly successful applications of scientific knowledge in systems where empirical regularities are
rare become intelligible. Indeed, it is only if
mechanisms operate in such ‘open systems’ that
successful lay interventions in nature in the form
of labour are possible.

A second sense of ’empirical’, which can be equally confusing, is ‘that which might be other than it
is’. Many interpretations conflate and confuse
(i) questions of contingency if and where it occurs
in the ontological domain (i.e. in the relations of
objects and events) with (ii) questions of contingency – or better ‘fallibility’ – in the epistemological domain (i.e. in the relation between the
world and our knowledge). (i) and (ii) are themselves only contingently related, and furthermore,
within (i) there is also the common non-sequitur
mentioned above in which it is assumed that because
it is logically possible that the world itself may
suddenly change, everything in our world is contingently related. Another source of confusion is
generated where the empirical is associated with that
which is referred to by logically contingent statements in contrast to the necessary truths of
analytic statements.

On our account, both contingency and necessity
characterise the real as a whole, and not just that
part which happens to be empirically observable:

much of the real could be ‘other than it is’, but
there are also natural necessltles. It is contingent
whether we know either case, but whichever is the
case has nothing to do with logical relations of
statements, for, as we saw earlier, necessity in the
world can be described by either logically necessary
or logically contingent statements.

The Theoretical
We have argued that theory makes its strongest
claims about [43] necessary relations in the world
and about the natures of the objects in virtue of
which they obtain. It does so by ‘anchoring itself’

upon abstract concepts, but these, on their own,
permit less committal statements about contingent
relations occurring in common concrete configurations.

The latter require ’empirical analysis’.

This interpretation had several corollaries:

(1) Given that theory makes claims about the real
and is not a heuristic aid for ordering a privileged
empirical knowledge, the relation between observation
and theory is not to be understood in terms of
correspondence rules but in terms of statements
about causal connexions between real objects.

(2) With the development of conventionalist
critiques of positivism and the renewed interest in
the history of science by philosophers of science,
it has often been noted that scientists sometimes put
more faith in their theories than in observations,
even where the latter appear to contradict the former.

This behaviour has sometimes been rationalised as a
healthy ‘tenacity’ which protects newly-emerging
theories from premature refutation. Given our agreement with the critique of theory-neutral, certain
observation, this is unobjectionable. But because
this critique fails to reject empiricism’s flat
ontology, it fails to note that the real disjunction
between mechanisms and events also gives scientists
good grounds for being sceptical about the significance of their non-correspondence.

(3) As they lack a concept of natural necessity
and a distinction between necessary (internal) and

contingent (external) relations, these philosophies
have difficulty sustaining any distinction between
empirical research and theoretical reflection.

(4) Just as there are no grounds for identifying
the concrete with the empirical, there are none for
identifying the abstract level of causal powers and
mechanisms with the unobservable as Keat and Urry
tend to do [44].

(5) What is theoretical has nothing to do with
difficulty or unfamiliarity. Commonsense or informal knowledge contains many implicit assumptions
about real necessities. Commonplaces – such as the
claims that ‘we must eat in order to survive’ or
‘we are all mortal’ can therefore be, in our sense,
as ‘theoretical’ as that knowledge of natural
necessities which is the product of considerable
scientific labour and which is usually exclusive to
a minority and unfamiliar to the masses. In saying
this, I am not trying to invest commonsense with any
privileged status; it is often content with ignorance about the nature of the objects in virtue of
which mechanisms operate, or else is mistaken about
them, as in the characteristic error of reification
of social relations and processes. I would simply
wish to deny that scientific and lay knowledges are
an incommensurable pair of autonomous ‘discourses’

about which a priori judgements can be made. Commonsense is certainly characteristically an ‘unexamined
discourse’, but if we try examining it, we can find
examples of interpenetration with scientific discourse. The adequacy of particular lay and scientific
knowledges is a substantive and not a philosophical
question. We cannot simply contrast an immaculately
conceived ‘science’ or ‘theory’, whose privileged
status is guaranteed by the conditions of its production, with ‘ideology’, which is similarly condemned
by its conditions of production. Therefore the
abstract concepts upon which we anchor our analysis
may include some which are quite mundane. As we
have seen, competing scientific ‘discourses’ at one
level (e.g. social theory) may share commitment or
indifference to concepts which are crucial at
another. Without a recognition of the stratification
of the real, such asymmetries tend to be interpreted
as evidence of the autonomy and incommensurability of
discourses or paradigms in relation to some flat
ontology. The fiction of incommensurability arises
from: (i) a flat ontology; (ii) an unawareness of the
hermeneutical character of discourse; (iii) a blindness to the mundane assumptions common to several
discourses produced by the reduction of a discourse
to those concepts which are unique to it; (iv) the
mistaken belief that discourses must be logically
continuous for translation between them to occur;
(v) a conception of networks of concepts composing
discourse existing in a kind of equilibrium, rather
than differential stress.

Critical implications
This discussion has important critical implications
for the way in which analyses and explanations of the
concrete are conducted. Neither empiricism nor
rationalism are of any help here; the former cannot
comprehend the role of theory, the latter cannot
grasp how theory has any purchase on the real. We
have seen that the false view that the move from
abstract to concrete is deductive and purely internal
and unique to marxist theory is based upon notions of
observation as entirely theory-determined, and discourses as entirely discrete and incommensurable.

These views legitimise a kind of reductionist or
‘pseudo-concrete’ analysis in which the concrete is
simply reduced to the abstract and in which the
extent of contingency in the systems of interest is
radically underestimaued. But the mediation of

discourses required by concrete analysis and the nondeductive relation between abstract and the concrete
need not be seen as problems. On the contrary, they
prevent a blinkered imprisonment within the ‘selfratifying circles’ [45] of the consecrated abstract
concepts of marxism. ‘Pseudo-concrete’ research
produces precisely that ‘forcing [of] the significance of certain events’ (Sartre), that ‘intellectual
agoraphobia’ (Thompson) or that ‘naming-of-parts
approach’ which has so often passed for analysis.

This is not to deny that theoretical reconstitutions
of higher-order abstract concepts are worthwhile,
only that they cannot provide the sole basis for the
generation of new concepts. These have to be
integrated into the existing networks of concepts if
they are to be meaningful and usable, and the integration usually involves meaning change in the network rather than a simple accretion of knowledge.

But the integration should not take place only ‘from
above’, in terms of Figure 1, but should also be
connected to more mundane and concrete concepts.

A one-sided integration has characterised much of
the recent writing on the state which has shown that
the hypotheses were already implicit in existing
theory, and only needed a theoretical exegesis, an
appropriate ‘reading’, to ‘draw them out’ [46]. It
is certainly important to realise that the state has
its own ‘labour processes’, that it is the ‘condensate of class struggle’ or an ‘instrument of the
dominant fractions of capital’, or whatever, but it
is also important to relate it to ‘lower-order’ concepts if these ideas are to inform concrete study.

And if these ‘lower-order’ concepts refer to some of
the same objects (differently understood, of course)
as bourgeois analyses e.g. ‘governments’, ‘civil
service’, this does not make the analysis irredeemably ’empiricist’. These ‘lower-order’ concepts are
certainly not ‘operationalisations’ of ‘the~retical
terms’ (which is how empiricists would see the
matter), but different aspects of the object of

Reductionism in Economic Analysis
One of the main forms of reductionism in marxist
‘economic’ analysis is an interpretation of empirical
patterns as simple manifestations of the abstractions
developed in Capital. A common tendency (fortunately
becoming rarer now) is the making of cavalier, unqualified assumptions about value movements on the
basis of price movements of physical volumes of plant.

Given that mechanisms and their effects rarely
correspond spontaneously in a one-to-one fashion,
this generates an actualist dilemma: either the
abstract tendencies do not exist as such or else
the empirical phenomena have to be distorted so that
they are made to reflect the abstract. This dilemma
is familiar in concrete studies of class but it has
also characterised neo-marxist analyses of uneven
development. In the latter case, at worst, ideal
type representations of contingent empirical patterns
of development (such as centre-periphery, metropolissatellite forms) are assumed to be the unique expressions of capitalist development. The actualist
dilemma is confronted when counter-examples of
peripheral ‘autonomous’ capitalist development are
pointed out: either the claims have to be retracted
or the existence of the exceptions denied. A
similar problem arises when certain novel empirical
forms (e.g. runaway industries, neo-Fordism) are
extrapolated and granted epochal significance as
unique manifestations of the latest phase of capitalist development. In both cases, both the empirical
‘rules’ and the exceptions are quite compatible
wi th the abstract propositions of Harx’ s Capital.

Much the same result is produced where the effects


of a necessary relation forming part of an open
system (i.e. one whose internal and external parameters are inconstant) are projected onto the whole
of the system. For example, it is common to note the
necessary contradiction of capital accumulation in
‘Newly Industrialised Countries’ in which the cheapness of the labour power both blocks as well as
assists accumulation because it restricts the size of
the market. While it is true that the low-paid
cannot generate much purchasing power, it is contingent whether there may be sufficient numbers of other
people in those countries who are affluent enough to
create an internal market. The fact that the latter
cannot be known in advance on the basis of knowledge
of abstract necessities creates traps for pseudoconcrete research [47].

This failure ,to acknowledge contingency in economic
systems is not only produced by an implicit concept
of an empirical world. Particularly in dependency
theory it also derives from a theorisation of tendencies or mechanisms which ignores much of the
marxist theory which explains how they are grounded.

The tendencies float uneasily and unconvincingly
between the abstract and the concrete, neither
grounded in the former nor engaging with the
latter [48].

Common to these approaches is the expectation of
a ‘theory of uneven development’ which pre-empts its
concrete form, and which is as misguided as the expectation of a ‘theory of ideology’ which specifies,
in advance, its content. Once again, abstraction
can only be expected to help explain the structures
or mechanisms which produce the concrete [49].

Mandel’s Late Capitalism [50] is certainly not in
this league, for it attempts the ambitious project of
explaining concrete developments in the world economy
through abstraction by reference to movements in
value. However, the success of this project is considerably hindered by his ambiguous usage of the term
‘abstract’ and by his empirical treatment of
‘tendencies’ .

‘From the standpoint of historical materialism,
“tendencies” which do not manifest themselves
materially and empirically [Are these terms meant
to be equivalent?] are not tendencies at all.

They are products of false consciousness, or for
those who dislike that phrase, of scientific
errors.’ [ 51 ]
Having excluded the possibility of a non-empirical
real world, he is then forced to regard abstract concepts which refer to it as having no explanatory
purchase on the concrete.

‘As soon as “laws of development” come to be
regarded as so abstract that they can no longer
explain the actual process of concrete history,
then the discovery of such tendencies ceases to
be an instrument for the revolutionary transformation of this process. All that remains is a
degenerate form of speculative socio-economic
philosophy in which the “laws of development”
have the same shadowy existence as Hegel’s “world
spirit” … ‘ [52]
Here, Mandel comes face-to-face with the idealist
alternative generated by the retention of an unstratified ontology of the empirical. He obviously
sees its unacceptable impl ications, and in orde,r to
avoid retracting laws of tendency he turns a blind
eye to the conspicuous absence of empirical regularities or simple empirical manifestations of laws of
tendency. In other words, his response to the
actualist dilemma is to ignore its existence. And
in the rest of the book, ‘tendencies’ appear to be
inferred from or read into empirical patterns exhibiting very little regularity [53]. But this is not
necessarily wrong for, given the non-identity of
mechanisms and their effects, empirical regularities
are neither necessary nor sufficient for retroducing

the existence of mechanisms.

An associated misconception about abstract/ concrete
and theoretical/empirical relations in marxist
‘economic’ theory is ‘deductivism’ [54]. Here
theory is supposed to provide a set of propositions
from which empirical forms can be logically deduced,
and it is assumed that this deduction provides an
explanation. From the statement ‘All capitalists
employ wage-labourers’, we can deduce that any
particular capitalist must employ wage-labourers,
but this does not explain why this is so. Therefore,
the point which has often been made in debates about
value-theory, that prices cannot be deduced (or,
which is the same thing, ‘calculated’) from values,
does not count as a legitimate argument against its
explanatory ability; it could still explain price
movements (though this is not its intention), the
origin of profit etc [55]. The theory could only be
expected to be formulable mathematically in such a
way that concrete movements were calculable if real
world causal processes happened to conform to the
relations of logic. However, the uneven relation
of use-value and exchange-value guarantees that this
cannot be so.

In theoretical discussions (e.g. of reproduction
formulae) we often abstract from this unevenness by
assuming a fixed relationship between use-value and
exchange-value, as Marx often did [56], but while
this may be a convenient heuristic aid, it cannot
possibly be used as a simplifying assumption in the
study of concrete development. Capital accumulation
in the face of the pressure of the law of value
depends upon a changing relationship between usevalues and exchange-values. In virtue of this:

‘There is, then, no necessary inner relation
between the value of the constant capital, nor,
therefore, between the value of the total capital
(~ c+v) and the surplus value.’ [57]
The element of contingency introduced by this unevenness is also ignored by those accounts of the tendency of the rate of profit to fall which turn the
contingent empirical questions of the relation betweeI
between technical composition and organic composition
into a priori ones.

Deductivism’s associated neglect of careful
initial description and conceptualisation is a
particularly common occupational hazard in mathematical analysis in marxian econonics, where conceptualisation is so often made the slave of quantification. The quantities easily become little more than
‘variables’ and ‘functions’ which take on a life of
their own, cut off from the theoretical setting of
Marx’s abstractions which exhaustively examine their
contexts and determinations [58]. And the matter is
made all the more complicated by the fact that
exactly the same kind of ‘abstraction’ from real
determinants actually underpins the concrete social
practices which produce exchange-value [59]. In
other words this misleading form of abstraction, this
chaotic but ‘practically-adequate’ conception, is
actually constitutive of marxism’s object.

In all these cases, the failure to acknowledge
contingent relations between the abstract and concrete generates a monism. If the mechanisms
abstracted can in fact lead to several different
concrete results, then the denial of this contingency will generate several competing monisms, each
able to cite (carefully selected) ’empirical evidence’. And the reductionist character of this kind
of analysis will also seriously underestimate the
degree of internal differentiation and flexibility
in its objects.

The political consequences of monism and

reductionism are a failure to grasp the complexities
of the concrete, whether they be the rigidities and
diverse forms of capital, which are so important for
understanding the crisis, or the web of crosscurrents which constitute the concrete forms of the
labour movement. For example, depending on which
monism you choose, this can lead either to an
unwarranted optimism about the potential of the
working class or a defeatist unfounded pessimism
produced by a projection of bad features of the
labour movement onto the whole. And this latter kind
of pessimism is in no small part reinforced by the
self-justifying and self-induced political isolation
of reductionist marxism.

The ideas in this paper owe much to the work of Roy
Bhaskar and Rom Harre – more than can be indicated by
mere references. I would like to thank Roy and also
Simon Duncan, Tony Fielding, Suzanne Mackenzie,
Peter Saunders, John Urry, Anthony Giddens, Roy
Edgley and Scott Meikle for comments on earlier
drafts. The usual disclaimers apply.








Search for a Method, 1963, Vintage, p.27; ‘studying facts’

now seems unsatisfactory, but I don’t think this vitiates
his criticism.

ibid. p.28, emphases in original. Also note the following
widely-quoted passage:

‘Val~ry is a petit bourgeois intellectual, no doubt
about it. But not every petit bourgeois intellectual
is Val~ry. The heuristic inadequacy of contemporary
Marxism is contained in these two sentences ….

Characterising Val~ry as a petit bourgeois and his work
as idealist, the Harxist will find in both alike only
what he has put there.’ (ibid, p.56)
Marxism and Literature, 1977, Oxford University Press,

The Poverty of Theory, Merlin, 1978, p.303. A daft title,
some polemical excesses, and a general failure to recognise
the importance of ‘structure’ (cf. P. Anderson, Arguments
Within English Marxism, NLB, 1980) should not be allowed
to detract from the importance of this critique as an
(albeit one-sided) corrective to some of the idealist
elements within Althusserianism.

ibid. p.346.

R. Bhaskar, 1975a, A Realist Theory of Science, Leeds Books;
1975b, ‘Two Philosophies of Science’, New Left Review 94;
1979, The Possibility of Naturalism, Harvester Press,
Brighton. R. Harr~, 1970, The Principles-of Scientific
Thinking, Macmillan; 1972, The Philosophies of Science,
Oxford UP, London. R. Harre and E.H. Hadden, 1975,
Causal Powers, Blackwell, Oxford. R. Harre and P.F.

Secord, 1972, The Explanation of Social Behaviour,
Blackwell, Oxford. R. Keat and J. Urry, 1975, Social
Theory as Science, Routledge and Kegan Paul.

P.K. Feyerabend, 1970, ‘Consolations for the Specialist’

in 1. Lakatos and A. Musgrave (eds.), Criticism and the
Growth of Knowledge, Cambridge UP.

Cf. A. Collier, 1979, ‘In Defence of Epistemology’,
Radical ghilosophy 20, pp.8-2l, and T. Skillen, 1979,
‘Discourse Fever: Post-marxist modes of production’,
Radical Philosophy 20, pp.3-8.

K. Marx, 1973, Grundrisse, Penguin, p.lOl.

ibid. p.lOO.

Bhaskar, 1975a, op.cit. pp.20l-02.

Harr~ and Madden, op.cit.

Harr~, 1970, op.cit.

Harre and Madden, op.cit., p.llO.

Bhaskar, 1975a, op.cit., p.95.

Collier, ibid.

Cf.’ ibid and M. Hesse, 1974, The Structure of Scientific
Inference, Macmillan.

Harre and Madden, op.cit.

ibid. p.80.

Bhaskar, 1975a, op.cit., p.20l.

It has often been said that the entire three volumes of
Capital are a definition of capital.

J. Banaji, 1977, ‘Modes of Production in a Materialist
Conception of History, Capital and Class 3, pp.1-44.

Bhaskar, 1979, op.cit., p.54.

Bhaskar~ 1975a, op.cit., p.52.

Bhaskar, 1975a, op.cit., pp.199-2l5.

Marx, 1857 Introduction, in Arthur, C. (ed.) The German
Ideology, Lawrence and Wishart, p.14l.

The translation in the Penguin/NLR edition of the
Grundrisse is possibly less committal: ‘ … the abstract
determinations lead towards a reproduction of the concrete
by way of thought.’ (Marx, 1973, op.cit., p~lOl). Compare
Althusser: ‘ … the process that produces the concreteknowledge takes place wholly in the theoretical practice.’

(L. Althusser, 1969, For Marx, NLB, p.186).

Marx, 1973, op.cit., p.lOl.

D. Sayer, 1979, Marx’s Method: Ideology~ Science and
Critique in Capital, Harvester Press, Brighton.

Bhaskar, 1975a, op.cit., p.l07. Compare Raymond Williams’

distinction between determinism and determination in his
1973b ‘Base and Superstructure in Harxist Cultural
Theory’, New Left Review 82, and Sartre’s concept of the
‘project’, op.cit., pp.9lff.

Bhaskar, 1975a, op.cit., pp.163ff. and his 1979, op.cit.,
pp. 124ff.













T.W. Adorno, 1976, ‘Sociology and Empirical Research’,
p.239 in P. Connerton (ed.), Critical Sociology, Penguin.

Marx, 1973, op.cit., p.96.

ibid. p.87.

ibid. p.88.


As Nicolaus notes, Marx was keenly aware of the limitations
of a purely idealist dialectic of categories: ‘ … as if
the task were the dialectic balancing of concepts, and not
the grasping of real relations’, Marx, 1973, ibid. pp.36
and 90.

Bhaskar, 1975a, p.149.

See the Postface to the 2nd edition of Volume 1 Capital
(1976, Penguin, p.l02):

‘Of course the method of presentation must differ in
form from that of inquiry. The latter has to appropriate the material in detail, to analyse its different
forms of development and to track down their inner
connection. Only after this work has been done successfully, if the life of the subject-matter is now reflected back in the ideas, then it may appear as if we have
before us an a priori construction.’

Bhaskar, 1975b, op.cit.

Bhaskar, 1975a, op.cit., pp.9lff.

Here, as in many places, we could insert the qualification
– ‘what are believed to bc’ – but given that all knowledge
isfaliible, there seems little point in making a special
emphasis of this here.

R. Keat and J. Urry, op.cit.

‘History and Theory’, in History Workshop 6, 1978, p.2.

Compare Sartre, op.cit., pp.27-28.

See the criticisms of Jacobson, D., Wickham, D. and
Wickham, J., 1979, review of Die Neue Internationale
Arbeitsteclung by F. Fr/jbel, J. Heinricks and O.K. Rowolt,
in Capital and Class 7, pp.125-30, and Nayyar, D., 1978,
‘Transnational Corporations and Manufactured Exports from
Poor Countries’, Economic Journal~ 88, pr.59-84.

I have also criticised Castells’ marxist analysis of
capitalist urbanization in The Urban Question, 1977, Arnold,
and City~ Class and Power, 1978, Macmillan, London, on
these grounds in my ‘Theory and Empirical Research in
Urban and Regional Political Economy’, 1979, University of
Sussex Urban and Regional Studies~ Working Paper, 14.

David Harvey develops this point in ‘The Geography of
Capitalist accumulation: a Reconstruction of the Marxian
theory’, Antipode 7 (2), pp.9-21. Cf. also E.O. Wright,
‘The Value Controversy and Social Research’, New Left
Review 116, 1979, which makes some points that are convergent with this.

E. Mandel, 1975, NLB.

ibid. p.20.


As would be expected of one caught in a dilemma, Handel’s
larger argument contains several bewildering oscillations
between different interpretations of the ‘abstract’, and
these perhaps account for his curious interpretations of
Capital, particularly the reproduction schemes of Vol.II.

Cf. Harre, 1970, op.cit.

B. Fine, 1980, Economic Theory and Ideology, Arnold.

Cf. also S. Heil~le, ‘Dialectical Contradiction and
Necessity’, in J. Mepham and D.-H. Ruben (eds.), 1979,
Issues in Marxist Philosophy: Vol.l~ Dialectics and Method,
Harvester, Brighton.

‘If now our spinner, by working for one hour, can convert
l23lbs. of cotton into l23lbs. of yarn, it follows that in
6 hours he will convert 10 lb. of cotton into 10 lbs. of
yarn.’ Marx, 1976, Capital, Vol.l, Penguin, p.297.

K. Marx, 1971, Capital, Vol.3, pp.46-47, Lawrence and

Cf. the important critiques of modes of abstrac60n in
economic analysis developed by Bettelheim in his critique
of Emmanuel in the latter’s Unequal Exchange, 1972, NLB,
pp.27lff. and by Haurice Dobb in his ‘The Trend of Modern
Economics’, 1937, reprinted in E.K. Hunt and J. Schwartz
(eds.), A Critique of Economic Theory, Penguin, 1972.

A. Sohn-Rethel, Intellectual and Manual Labour, 1978,


Download the PDFBuy the latest issue