David Harvey: Explanation in geography

Ron Johnston1 School of Geographical Sciences,

This has been prepared for a book on ‘Key Texts in

NOT TO BE CITED WITHOUT THE AUTHOR’S PERMISSION

The 1960s was a turbulent decade in geography: by its end, the discipline incorporated practices very different from those deployed ten years earlier. Building on discontent with disciplinary practices born of experience of interaction with scholars from other disciplines in the American Office of Strategic Services during World War II (Barnes and Farrish, 2006), a number of geographers – notably a cohesive group of faculty and graduate students at the University of Washington, Seattle – began in the mid-1950s to promote a very different vision of geography (both physical and human) based on the ‘scientific methods’ deployed by physicists and, in the social sciences, economists. By the early 1960s, this group was rapidly attracting adherents at a number of major US graduate schools, and in 1963 one observer-participant claimed that the ‘revolution’ promulgated from Seattle was successfully over.

Almost contemporaneously, a similar ‘revolution’ was taking shape in the UK, based at the University of Cambridge. Two recently-appointed faculty members – Dick Chorley and – were attracted to the ‘scientific model’ (which Chorley had encountered and adopted while a geology graduate student in the USA), and they began to teach related material, mainly statistical methods, in their introductory practical classes for geography undergraduates. One of the demonstrators in the first year for that course (in 1960) was a postgraduate student working in historical geography – David Harvey.

Although the shift in geographical practices these two groups were advancing is often referred to as the discipline’s ‘quantitative revolution’, it involved much more than just applying mathematical and statistical techniques to geographical data. It was fundamentally a ‘theoretical revolution’, which changed the entire mind-set of how research was to be undertaken and new knowledge presented. Words such as ‘theory’, ‘model’, ‘hypothesis’ and ‘law’ become common in the geographical lexicon, as researchers strove to produce knowledge that was cumulative in the sense generally appreciated by physical scientists, building on earlier research to advance their ability not only to explain the world that they observed but also to modify (improve) it.

1 My thanks to Les Hepple, Tony Hoare, Kelvyn Jones, Charles Pattie and Eric Pawson for valuable discussions of this essay and comments on draft versions. When a discipline is experiencing such major change, new teaching materials are needed to introduce students to (and justify) the ‘revolutionary’ practices. Introductory textbooks usually lag behind such changes – publishers have to be convinced that there is a viable market for ‘revolutionary tracts’! This was so with the 1960s ‘new geography’. The first books clearly enunciating the changes appeared in 1965 (Haggett’s Locational analysis in human geography – this was very much a publisher’s gamble on the shape of the discipline’s future – and Chorley and Haggett’s edited volume Frontiers in geographical teaching: Bunge’s, 1962, monograph on Theoretical geography had only a limited circulation, although a revised edition in 1966 attracted wider attention). A spate of texts oriented towards the ‘revolution’ only emerged some five years later. It included David Harvey’s Explanation in geography, the product of almost a decade teaching undergraduates at the University of Bristol about the ‘new’ scientific basis to geographical work, building on discussions with colleagues in Sweden and the USA (where he spent considerable time during the decade); as he puts it in the Preface, writing the book was part of his learning experience as he developed teaching materials.

Harvey’s book – like that of the other ‘revolutionaries’ he joined – reflected a deep dissatisfaction with geographical practices experienced when he was an undergraduate and in his interactions with researchers immediately thereafter. For him, too, quantification was necessary but far from sufficient: measurement was a requisite tool, but much more important was for both human and physical geographers to appreciate and deploy the philosophy which under-pinned the ‘fantastic power’ of the scientific model. Hence he explored – and wrote a book (he termed it an ‘interim report’) about – ‘the ways in which geographical understanding and knowledge can be acquired and the standards of rational argument and inference that are necessary to ensure that the process is reasonable’ (p. viii). This exploration took him into a literature previously almost entirely ignored by geographers, so that although he includes plentiful references to a small number of geographers (especially those instrumental in fomenting the ‘revolution’, but also two – Hartshorne and Sauer – whose views he largely opposed) his ‘Author index’ indicates how heavily he drew on philosophers of science – Ackoff, Braithwaite, Carnap, Churchman, Hempel, Kuhn, Nagel, Popper (to a lesser extent; falsification is very summarily dismissed) and even Einstein and Russell (although neither Ayer nor Wittgenstein is mentioned) – as well as mathematicians/statisticians such as Anscombe, Blalock, Fishburn, Fisher, Krumbein, Sneath, and Sokal.

From the outset, Harvey focused on explanation. After a brief introductory chapter setting out his main concern as being with methodologies – with how geographers should produce explanations – rather than philosophies of science (despite their foundational role), he moved to a discussion of what explanation means. It is defined as ‘making an unexpected outcome an expected outcome, of making a curious event seem natural or normal’ (p.13) because it can be shown to be generated by similar processes and in similar conditions as previous events of the same type. Harvey’s concern was with ‘rational explanation’, statements verifiable by others because the procedures involved in their production can be repeated and/or are open to scrutiny. It is a ‘formal procedure … [the] hard inner core of methodology’ (p.23).

Following that brief framework-setting introduction (26 pages only), the remainder of the book comprised five main sections dealing with: explanation in geography;

2 theories, laws and models; languages; ‘models for description in geography’; and ‘models for explanation in geography’. At the outset he contrasted (in an oft-reprinted diagram) the inductive or ‘Baconian’ path to explanation with his preferred deductive route, which proceeds through the establishment of a model representing the researcher’s image of the world, the derivation of hypotheses regarding some aspect of that image, testing the hypothesis’s validity, and the formulation of theories and laws synthesising the knowledge gained – from which revised models can be derived.

Four terms/concepts stood out in the lexicon associated with this route to explanation. Hypotheses were presented as logically consistent ‘controlled speculations’. They can never be tested absolutely – conclusions are always provisional – but they guide the production of scientific knowledge in a rigorous (and hence replicable) way.

The key outcomes of testing hypotheses are laws and theories. Laws are sometimes presented as universal truths – statements whose validity is constrained by neither time nor space. It is never possible to reach such conclusions, but scientists act as if they can, using their findings as statements of the current ‘conventional wisdom’ encapsulating what we already know and providing the foundations for further scientific exploration. They are rigorously produced provisional conclusions representing the contemporary state of knowledge. For geography (especially human geography, although much physical geography at the time – such as the Davisian model of landscape evolution – similarly lacked rigorous underpinnings), they contrasted sharply with the ‘explanatory sketches’ traditionally offered as accounts of ‘observed reality’. The production of laws in geography, according to Harvey, involved searching for ‘hidden order within chaos’: because the results of such searches are likely to be provisional/tentative, Harvey suggested that geographers might prefer the concept of ‘law-like statements’, general claims that are both ‘reasonable with respect to experience and consistent with respect to each other’ – a coherent body of knowledge corresponding with the observed ‘reality’.

Theories are systems of linked statements about defined subject matter. These may be entirely closed systems – as with Euclidean geometry; they may be sets of deductive statements derived from accepted axioms; or they may be less-formally stated ‘sketches’. They are not just speculative ideas – as sometimes implied in vernacular uses of the term: anyone can fabricate such a ‘system of apparent wisdom in the folly of hypothetical delusion’ (p.97: Harvey is quoting James Hutton via Chorley). Scientific theories take ‘such speculations and transform ... them from badly understood and uncomfortable intrusions upon our powers of ‘pure’ objective description into highly articulate systems of statements of enormous explanatory power’ (pp. 87-88).

Various types of theory extend along a continuum from highly-formalised, internally- closed sets of statements (as with many forms of mathematics and logic), through sets of statements which are only partial (either because their primitive terms, or axioms – the assumptions on which they are built – are incomplete or because the deductions from those foundations are not fully elaborated), to what Harvey terms ‘non-formal theories … statements made with theoretical intention, but for which no theoretical language has been developed’ (p.98). The last type ‘scarcely conform in any respect to the standards of scientific theory’ (p.130), and are characteristic of previous ‘theoretical’ work within geography. Geographers had to move forward, he argued,

3 either by deriving theories from axiomatic statements – more likely to be feasible in physical geography, which can deduce, for example, landscape-producing processes and the likely resultant forms from physical laws – or, in human geography, by generating assumptions about human behaviour from which statements about spatial patterns can be deduced.

Whatever the theory’s origin, establishing its empirical status involves moving to the final key word in the new lexicon Harvey was interpreting for geographers – the model. This has a diversity of meanings in both popular and scientific language; for Harvey’s ‘new geography’ a model was a representation of a theory – i.e. an outcome of a series of law-like statements. Such representations became the source for hypotheses, leading to tests of the theory’s empirical validity.

An example of a (human) geographical theory illustrating these fundamental concepts is central place theory: it underpinned a great deal of geographical work in the 1950s- 1960s. Harvey showed that it was derived from a set of fundamental economic postulates (assumed laws) about consumer and provider behaviour (i.e. profit- maximization for providers and minimization of travel costs for consumers) and the nature of the goods/services being supplied/demanded. These are linked in a single theory from which it is possible to deduce the spatial arrangement of service centres. Models could then be derived showing the expected morphology of that spatial arrangement in different contexts, which had common features such as the hexagonal arrangement of centres in nesting hierarchies. Specific hypotheses could then be tested in particular empirical situations.

For Harvey, the most important of these concepts was theory: without theories ‘the explanation and cognitive description of geographic events is inconceivable’ (p.169). But how could such theories be expressed? The language to be deployed was that of mathematics – ‘the language of science’ – within which he concentrated on two sub- fields: geometry as the language of spatial form (geography being defined as the study of ‘objects and events in space’ – p. 191); and probability as the language of chance, necessarily used because ‘the world is governed by immutable chance processes’ (p. 260) so that precise prediction is rarely possible, especially so given the extent of our ignorance about those processes. A scientific geography will not be a deterministic geography, therefore, but rather comprise probabilistic statements of likely explanations (of the ‘hidden order within chaos’), hence the use of statistics in evaluating hypotheses.

Given the key components of the scientific method and its language, as applied to geography (both human and physical; Harvey saw no difference between the two in their methodological structure), Harvey then dedicated two sections of the book to modelling in geography – descriptive and explanatory. These are, in effect, chapters about methods: the first deals with measurement and how one portrays the world – how information is collected, classified and displayed; the second with procedures for testing hypotheses of cause and effect.

In the concluding chapter, Harvey summarised 480 pages of detailed material as ‘some rough and ready guidelines for the conduct of empirical research in geography’, presenting the tools that might be used when we ‘have to pin down our speculations, separate fact from fancy, science from science fiction’ (p.481). That is

4 done scientifically by producing ‘an adequate corpus of geographic theory’ – coherent statements about aspects of the world which identify what knowledge geographers produce, validated by their adoption of the protocols of scientific method. For him, geography in the 1960s lacked such a clear identity and sense of direction, hence the clarion call in the final pages of his text (p.486): Without theory we cannot hope for controlled, consistent and rational explanation of events. Without theory we can scarcely claim to know our own identity. … theory construction on a broad and imaginative scale must be our first priority in the coming decade. … Perhaps the slogan we should pin upon our study walls for the 1970s ought to read: ‘By our theories you shall know us’.

Explanation is a long and detailed book. It is not difficult to read once the basic concepts are appreciated, but it is not presented as a textbook – or at least not as such books are now presented: there are suggested readings at the end of each chapter, but no boxes or other devices designed to focus on key ideas and/or exemplars. Harvey never intended to write such a textbook: as he expressed it ‘The book was written for anyone who cares to read it and for anyone who cares to use it in whatever way or ways they find congenial or useful. … The sooner we stop writing for “an” audience … or for “the beginning graduate student” the better off we will be’ (Harvey, 1971, 323). His goal was nevertheless that of other textbooks: to introduce the methods geographers (should) deploy to produce knowledge, beginning with the key methodological protocols and then setting out detailed procedures. It is thus similar to other contemporary volumes, notably Spatial organization: the geographer’s view of the world (Abler, Adams, and Gould, 1971), which began with discussions of ‘scientific method’ and research procedures but then, in a much more ‘student- friendly’ way, illustrated these with detailed (human) geographical examples; Haggett’s (1965) Locational analysis in human geography, put procedures after examples – they became separate books in the second edition – but said very little about ‘scientific method’. Earlier books, such as Gregory’s (1963) Statistical methods and the geographer, were entirely about procedures. (Gregory’s book is not referenced in Explanation.)

Explanation’s accessibility – at least to Harvey’s peers – was testified by reviewers. They saw it as important, especially so as it was the first full-length treatment of what was involved in geographers adopting the methodology (and underlying philosophy) of the natural sciences – what geographers later referred to with increasing frequency as positivism (the term is not in Explanation’s index!). But its impact was probably less than its originality and depth suggest, given that it was the first full statement of what a ‘scientific geography’ should look like.

Explanation is unlikely to have been widely used as a textbook – certainly so for undergraduates – for three main reasons. The first is its density and presentation. Many in the early 1970s will have been directed to it, as supplementary reading for the increasingly-popular courses on the history and philosophy of geography – but it was too detailed to be deployed as the main text for a course on geographical methods. (Harvey’s course at Bristol was probably a rarity within undergraduate programmes and the book appeared more than a decade before postgraduate courses were instituted in British geography departments: it may have attracted more attention in US graduate schools, but would have competed with more ‘accessible’ books such

5 as Abler, Adams and Gould, 1971.) Some undergraduates will have been invited to read particular parts as the basis for tutorial/seminar discussions; others may have either discovered it serendipitously on a library/bookshop shelf or been directed towards it, and become absorbed by its material. But for most students (and also academics intrigued by and attracted to the changes in their discipline) its main role was as a reference text, something to be used when seeking detailed material about, for example, the key concepts of theory, law, hypothesis and model. It paid too little attention to the examples that are so often the key to student appreciation of a set of ideas, and so loomed in the background: certainly its chapters on procedures did not take the ‘how-to’ form typical of introductory textbooks. (Amedeo and Golledge’s later, 1975, An introduction to scientific reasoning in geography was a more ‘user- friendly’ introduction to the protocols.)

Secondly, although original in its depth and catholicity Explanation was not entirely novel. ‘The shock of the new’ had hit geographers some years before with the publication of books such as Frontiers in geographical teaching (Chorley and Haggett, 1965), Locational analysis in human geography (Haggett, 1965) and Models in geography (Chorley and Haggett, 1967). These were much more instrumental in bringing the fomenting revolution in geographical practices to academic geographers’ attention (especially in the UK). For those converted, Explanation provided the detailed exposition needed to make them fully aware of the complexity of the practices they were planning to adopt. But, as Harvey made clear, the book was not about philosophy, only methodology within a particular philosophy, and although he had clearly addressed many philosophical issues in his wide reading it was later authors (notably Gregory, 1978) who explored issues of epistemology and ontology more fully – including those of the positivism Harvey embraced – and linked geographers to a much wider set of debates.

Thirdly, by the time Explanation appeared the practices that it advanced were being strongly contested – not least by David Harvey himself! Thus when he responded to by far the longest review/critique to appear – by Stephen Gale (1971), who thought it ambitious and stimulating, but deficient, because it is partial and to some extent incoherent, as both textbook and reference volume – Harvey had moved away, not from science but from that particular scientific philosophy and form of scientific method. As detailed in a number of early 1970s’ essays (reprinted in Harvey, 1973), he turned to Marxism as his source of theoretical appreciation, claiming that the scientific method he had previously advanced was ideological in that it sustained the political status quo (hence Harvey, 1974) and, however sophisticated its descriptions of the world, was unable to appreciate the underlying processes which produced them. Harvey certainly did not abandon theory: rather, he moved his theoretical stance, adopting a new set of protocols and procedures. As Peet (1998) put it, with Explanation Harvey proved to geographers their need for theory, but almost immediately realised it was the wrong theory: his clarion call remained intact – but the theories did not.

Although there was a clear ‘revolution’ in Harvey’s own approach to human geography almost immediately after Explanation was published, quantitative work remains a strong strand within the discipline – increasingly sophisticated in both its methods and the technology on which it depends. But that work increasingly distanced itself from its positivist foundations – particularly with regard to the search

6 for law-like statements. As discussed in recent essays (such as Fotheringham, 2006), most quantitative work involves the rigorous interrogation of large, spatially- structured data sets with the goal of finding order within a highly complex world but without implying that such order is fixed. It seeks to accrue ‘sufficient evidence on which to base a judgement about reality that most reasonable people would find acceptable’ (Fotheringham, 2006, 241), in contexts where understanding calls for the deployment of large, aggregate data sets. For such enterprises, much of the detail in Explanation is irrelevant whereas many of the methods it promotes have been superseded. Explanation was one of the foundation stones in the creation of a ‘new theoretical geography’, but the superstructure has since been substantially reconstructed.

The major change in Harvey’s work came in the early 1970s (as Harvey, 2002, 2006, illuminated), and he remains one of geography’s most influential scholars. He has also – uniquely – been the subject of two critical evaluations. The first (Paterson, 1985) covered the Explanation years and the first decade of Harvey’s Marxist explorations, but gave less than one-third of its space to the former and, while documenting the major discontinuity in his thought, gave little attention to what was identified as a major continuity – ‘the importance of general theory’. (This misrepresents Harvey for whom theory is the continuity, not ‘general theory’, whatever that might be.) The second (Castree and Harvey, 2006) appeared some 35 years after the switch in Harvey’s theoretical orientation and, given his productivity and the seminal nature of much that he has published since, not surprisingly pays relatively little attention to the first decade or so or Harvey’s career. Barnes puts that ‘first life’ into its social context and stresses its continuities with Harvey’s post-1970 output – commitments to geography, to politics (i.e. the application of geographical knowledge), and ‘perhaps most germane…, to theory’ (p.42); Gregory unpacks the lacunae in Explanation – the black-box system diagrams that expressed an ignorance of process and the sterility of mathematical language; and Sheppard identifies several more continuities, notably a concern for space and time – and space-time. But Harvey the Marxist scholar gets the bulk of the attention (as in Castree, 2004).

To paraphrase a statement made in another context, Explanation is undoubtedly now ‘more revered than read’ – and may always have been so. As Castree (2004, p. 181) puts it, Explanation ‘gave Harvey’s generation of geographers a heavyweight justification and manifesto for their project’, aligned ‘the discipline with the so-called ‘real’ sciences like physics and, for some geographers, boosted the discipline’s self- image’ – although many human geographers sought status within the social sciences rather than suffered from ‘physics envy’. But Harvey abandoned his generation – or many of them – for an alternative project, to which he attracted a new generation of converts. Indeed, according to his autobiographical essay, in some ways he abandoned the first project long before he completed it – having a ‘lust to wander and diverge, to challenge authority, to get off the beaten path of knowledge into something different, to explore the wild recesses of the imagination as well as of the world’ (Harvey, 2002, p.167). He did finish it, however, but responded to Stephen Gale’s (1971) review by saying that he was at a disadvantage because Gale had read the book and ‘I have never read it. What is more, I have no intention of doing so now’. Explanation was behind him, but remains a permanent and potent reminder of a crucial decade in geography’s turbulent recent history; extremely influential when published, as not only a pioneering exploration of ‘scientific method’ and its philosophical

7 underpinnings but also one of the first substantive geographical engagements with the social sciences.

Abler, R., Adams, J. and Gould, P. (1971) Spatial Organization: The Geographer’s View of the World. Englewood Cliffs NJ: Prentice-Hall.

Amedeo, D. and Golledge, R. G. (1975) An Introduction to Scientific Reasoning in Geography. New York: John Wiley.

Barnes, T. J. and Farrish, M. (2006) Between regions: science, militarism, and American geography from World War to Cold War. Annals of the Association of American Geographers 96:

Bunge, W. (1962) Theoretical Geography (second edition 1966). Lund: C. W. K. Gleerup

Castree, N. (2004) David Harvey. In P. Hubbard, R. Kitchin and G. Valentine, editors, Key Thinkers on Space and Place. London: Sage Publications, 181-188.

Castree, N. and Gregory, D. editors (2006) David Harvey: a Critical Reader. Oxford: Blackwell Publishing.

Chorley, R. J. and Haggett, P., editors (1965) Frontiers in Geographical Teaching. London: Methuen.

Chorley, R. J. and Haggett, P., editors (1967) Models in Geography. London: Methuen.

Fotheringham, A. S. (2006) Quantification, evidence and positivism. In S. Aitken and G. Valentine, editors, Approaches to Human Geography. London: Sage, 237- 250.

Gale, S. (1971) On the heterodoxy of explanation: a review of David Harvey’s Explanation in Geography. Geographical Analysis 3, 285-322.

Golledge, R. G.. (2006) Philosophical bases of behavioural work in geography. In S. Aitken and G. Valentine, editors, Approaches to Human Geography. London: Sage, 75-85.

Gregory, D. (1978) Ideology, Science and Human Geography. London: Hutchinson.

Gregory, S. (1963) Statistical Methods and the Geographer. London: Longman.

Haggett, P. (1965) Locational Analysis in Human Geography. London: Edward Arnold.

8 Harvey, D. (1971) On obfuscation in geography: a comment of Gale’s heterodoxy. Geographical Analysis 3, 323-330.

Harvey, D. (1973) Social Justice and the City. London: Edward Arnold.

Harvey, D. (1974) What kind of geography for what kind of public policy. Transactions, Institute of British Geographers 63, 18-24.

Harvey, D. (2002) Memories and desires. In P. R. Gould and F. R. Pitts, editors, Geographical Voices: Fourteen Autobiographical Essays. Syracuse: Syracuse University Press, 149-188.

Harvey, D. (2006) Memories and desires. In S. Aitken and G. Valentine, editors, Approaches to Human Geography. London: Sage, 184-190.

Paterson, J. L. (1985) David Harvey’s Geography. London: Croom Helm.

Peet, R. (1998) Modern Geographical Thought. Oxford: Blackwell.

9