History In Focus logo and homepage link

History in Focus

the guide to historical resources • Issue 2: What is history? •

What is history?

Book cover: 'The new nature of history'

Book Review


The New Nature of History: Knowledge, Evidence, Language

Arthur Marwick
Palgrave, 2001 pp. xvi, 334

Alun Munslow

University of Staffordshire

The British social historian Arthur Marwick, the author of well over a dozen books and many articles is probably best known for The Nature of History first published in 1970. The book has had two re-prints in 1981 and 1989. This latest incarnation is, as he says, 'totally recast and re-written' with the aim of giving it 'a new coherence and a new dynamic thrust' (xiii). This is predicated on his exploration of what he perceives to be the three fundamentals of doing history: knowledge, evidence and language. Hence it is we have the new title of what is virtually a new book The New Nature of History: Knowledge, Evidence, Language.

Professor Marwick wastes no time telling us what he is about and he makes no bones about his qualifications for writing it. On page 3 he says while he has no mandate to speak on behalf of historians generally, he believes he has 'thought longer and harder about the issues implicit in [his book] than most other members of the profession' (3). But even earlier, in the Preface, he has explained his own empirical-analytical take on doing history, namely that historical knowledge 'depends on highly skilled and difficult work among the primary sources' (xiii). The rest of the book amplifies this 'source based' position through some careful and considered argument but also, and this is as unfortunate as it is needless, many descents into wanton professional abuse. I shall divide my review into sections each reflecting upon the three fundamentals of doing history as seen from the Marwickian perspective.


Professor Marwick's first major claim is that the process of producing historical knowledge is a source-based enterprise. This is founded on his epistemological position, that of empiricism. But he immediately attempts to shift and obscure this suggesting he is too sophisticated to suggest historians try to get back to or even reconstruct past reality. As he says 'historians do not, as too many of my colleagues keep mindlessly repeating 'reconstruct' or even 'represent' the past' (xiii). Possibly being the 'mindless' historian who first argued Professor Marwick is a reconstructionist (Munslow Deconstructing History 1997: 18 and again in Munslow The Routledge Companion to Historical Studies 2000: 6, 59, 152, 184) and as I think the past can only be taken hold of as that representation (technically as a substitution rather than a resemblance representation) we call history, I feel I have an another task in reviewing this book. I shall also attempt - very briefly - to offer something of a narrativist and linguistically informed balancing argument to that of Professor Marwick's epistemological privileging of empiricism.

I should clarify for new readers what I mean by a reconstructionist approach to historical thinking and practice given that I think Professor Marwick qualifies. The touchstone of such an epistemological position is the assumption that the more carefully we study our sources and the more refined are our inferences the closer we can get to fulfilling Ranke's famous dictum of knowing either the past as it actually happened or, at the very least, coming pretty close to an understanding of what it really means. Reconstructionism thus accepts the correspondence theory of truth, which, briefly, accepts the match between the word and the world.

The central operational principle of reconstructionism is an antipathy to any kind of philosophising and especially the process of (testing) preconceived theories of explanation (which is a social science approach). The corollary, as I have suggested elsewhere it must be, is a belief in the complex but ultimately graspable meaning of the past that lies dormant in the sources (in Marwick's preferred descriptions of primary and secondary sources, and testimony that is witting and unwitting). Hence for reconstructionists any form of a priorism is anathema as it smacks of the subjective intruding into the objective (i.e., source based) gathering of historical knowledge (see the next paragraph for my further thoughts on this). It follows for reconstructionists that history is not a social science and is actually diminished by any theory. For the reconstructionist the very nature of the-past-as-history (unique events and individual people making decisions and taking actions) precludes the possibility of any viable prior theoretical understandings as the basis for historical knowledge. Consequently, the only foundations for historical knowledge are empiricist or data based methods and principles upon which reliable inferences about the meaning of the past can be drawn. Clearly misguided are those who would a priori construct the past (most notably and dangerously in Marwick's view those who are adherents of Marxism) or just as bad or worse, those who would ask awkward questions about truth, meaning, representation and narrative (those who I have called deconstructionists and Marwick calls postmodernists and who are, in his view, 'inheritors of Marxist radicalism' (241)).

Given Marwick's investment in the sources it is regrettable that he makes the odd slip in this respect. His empiricist errors range from the relatively trivial incorrect spelling of my first name in the index (p. 324) to his more significant confusion over the secondary sources in which he himself is referenced, and much worse, even though he refers to his sources, he can't distinguish between two completely different people. On page 303 for example he claims there are many references to himself in Jenkins' On 'What is History?' but there is actually only one, then he says there is no reading in it by him thus confusing that book with another by Jenkins The Postmodern History Reader. But what is even more distressing for the source based Marwick is that he gets confused between two Keith Jenkins. The Keith Jenkins he refers to on page 257 (just after he says my 1992 book Discourse and Culture: The Creation of America 1870-1920 is a series of biographical essays and not a structured monograph) is not the Keith Jenkins he means. He says 'Keith Jenkins, before abandoning Christianity for postmodernism, wrote a British Council of Churches pamphlet, The Closed Door: A Christian Critique of Britain's Immigration Policies (1984)' ( 257). Well, he didn't. It is a different Keith Jenkins. In failing to check his sources properly his inference is wrong and so, therefore, is his history.

As these empirical errors demonstrate Professor Marwick is a reconstructionist. He begins by restating the epistemologically obvious, that historians produce knowledge (though not always accurately as he admits and he himself demonstrates). Unfortunately, Professor Marwick goes on to gloss this rather unexceptionable insight with an attack on what he calls the auteur theory of history, notably the (supposedly postmodern) view that historians are all too often seen (or some of them at any rate) as 'individual literary figures, akin to best-selling novelists' (xiii). Quite why he wishes to characterise what he believes to be a particular approach (postmodern?) to history in this way I can't understand. I can only assume he objects to the recognition (and what, importantly in epistemological terms, flows from it) that historians are the creators of the texts they write. It seems Professor Marwick believes in the consensus theory of history at this point, implying that as a profession we collectively (rather than individually) move toward the truth (although correspondence lurks just off-stage).

Anyway, the existence of auteurs, I think he thinks, makes history less objective as a discipline (38). My reading of this is that he endorses the notion, rejected by Kant and most thinkers on history subsequently, that it is possible through the archive, to get back to the-past-thing-in-itself. If, for some reason (usually associated with personal beliefs or ontological existence?), we are unable to do this, I presume he thinks we must descend into subjectivity. In spite of subjectivity being the normal state of criticism and knowledge creation in the humanities and being unavoidable at the most fundamental of epistemological levels, subjectivity is clearly a bad thing. Fortunately for reconstructionists like Marwick it can be overcome through an empiricist methodology, which is, for sensible (sense-able?) historians, the only game in town. As you can imagine, from my subjective perspective as co-editor of the Routledge history journal Rethinking History: The Journal of Theory and Practice, I do not agree with that particular reconstructionist empiricist fiction.

The problem for Professor Marwick's version of doing history, of course, is that we cannot get back to the-past-thing-in-itself (res gestae), as surely he must realise? This is as unknowable as Kant's noumenal world. I recommend the whole retinue of reconstructionists (ranging from Geoffrey Elton, Gertrude Himmelfarb, Jack Hexter, Elizabeth Fox-Genovese to Elisabeth Lasch-Quinn and the contributors to their recent book Reconstructing History upon which Marwick heaps much praise) and Professor Marwick, read Kant's Critique of Pure Reason. This might dissuade them (though I'm not optimistic) from trying to find out what actually happened as if this were the only way to learn important and morally useful things about events, people and their actions in the past (1). All historical knowledge is mediated by the positions we adopt, the decisions we make about sources, who and what we are as human beings, our preferred arguments through which we connect events, the nature of our inferences, the manner in which we construct our narratives, who the history we write is for, our publishing constraints, and the theories we invoke to refer to past reality. A much better and elegant defence of objective knowledge and rational action theory (though such a defence is ultimately also as unconvincing to me as the crude tub-thumping of Professor Marwick) is Mark Bevir's The Logic of the History of Ideas (1999). I would recommend after reading Kant Professor Marwick read this book as it might offer him something of a stronger philosophical basis for his pre-judgements. But to get back to the plot of my review/narrative-linguistic defence of a broader conception of the nature of history. To recognise the ineluctable subjectivity of, and in doing history, does not diminish its importance as a cultural discourse or practice. Without being flippant, recognising the subjectivity of history as much as its empirical basis, means we are actually being more realistic (if you want to talk in terms of realism) about what and how we do history.

To argue, as Professor Marwick does, that the sources appropriately corroborated and verified are the only way we can grasp the meaning of the past in my view seriously diminishes history's cultural potentialities (well, in our Western culture) and our present constructive engagement with the past. It is not to deny the reality of the past or the experiences of people in it to recognise and work within the past's essential unknowability when cast in the form of history. There is another important point to be made here and it is one upon which Marwick expends much spleen. The essential unknowability of the past is only in part due to the failings of the empirical-analytical approach to doing history; it is also the result of the fact that historians compose history as a narrative. Whether Professor Marwick likes it or not, it is historians who write history (we are all auteurs now?) and this has serious implications for the creation of historical knowledge. Neither his effort at breaking the sources down into thirteen different types, nor his personal hierarchy of four explanatory factors are substitutes for some serious thinking about the un-knowability of past reality and how we come to terms as historians with its sublime nature. I am not suggesting Professor Marwick can't think seriously. It's just that he doesn't seem to want to.

Let's examine an example of what I take to be one of the occasional examples of his serious thinking. At one point he suggests (quite correctly and so far so good) that to write history historians have to devise a structure for their work and he gives two examples (206-214). Now, being a reconstructionist who is happy with the notion that content determines form (a key empiricist assumption for doing history), he believes a history book's structure reflects (and should properly reflect) the true nature of the past as much as can be determined according to the sources. As he says, the history as 'represented' (Note: hang on, a bit, I thought he said he didn't represent the past on p. xiii?) in the table of contents 'is a form governed by the fruits of work done in the primary and secondary sources' (207). It has, as he says 'nothing to do with the nonsense about the need for 'emplotment' or 'narrative' (in the imagined sense of the postmodernists)' (ibid). The fundamental point Marwick seems to be making here is that the structure in the history book is not imposed arbitrarily. Rather, it is derivative because, as he says, there must be 'a logical order, a series of connections and interrelationships (in short a 'structure') which will be as true to the actual aspects of the past...as it is possible for a historian producing knowledge about these aspects of the past to make it' (208). Marwick seems to be suggesting there is a story back there, which is full of meaning, and the historian can find it in the sources and replicate it (or, as I say reconstruct it) in the book?

While never denying the real and the need to always aim for truth (however truth is defined conceptually - see below), it was the American historian, and critic of crude empiricism, Hayden White who confronted the belief that the meaning of the past (usually couched as the most likely meaning translated into the truth conditional statement) can be read off from the sources and represented with the least amount of narrative imagination as the historical text. This is usually called getting the story straight and as a process it echoes the position of realist narrativist philosophers like W.B. Gallie and (in a rather more sophisticated fashion) Michael C. Lemon. Such a judgement is only sustainable, of course, if the historian accepts the empiricist assumption that content (the past) always determines form (history). White reasoned that historians ought to question the empiricist a priori that underpins Marwick's belief, first expressed in the 1970 edition of The Nature of History, that through advances in methodology and by standing on the shoulders of earlier generations of historians, there can be an absolute advance in the truthfulness of history (Marwick 1970: 21). This is a sentiment that clearly remains central to the latest edition of the book.

This epistemological view, I think, has few adherents in today's world of highly sophisticated constructionist history writing. This is in part due to White's questioning of the idea that the story of the past exists in the sources, and that the more we compare and contrast them the closer we can get to that story. In other words, White challenged the empiricist belief that the historical narrative is ultimately determined by the sources alone. White's position is that, as knowledge of the sources cannot produce the truthful interpretation (a 'truthful interpretation' is, as Keith Jenkins regularly points out, an oxymoron), then the knowable narrative cannot pre-exist in the events. This means it must be made available by the historian. White thus freed himself of the correspondence theory of truth to which Marwick is in debt, except when he chooses to attack auteurs and thereby, presumably, endorse the consensus theory of truth. I agree with Elizabeth Deeds Ermarth that the challenges to the modernist notion of objectivity are too extensive now to be dismissed as the ravings of a few mindless pseudo-historians as Marwick suggests (E.D. Ermarth 'Beyond History' Rethinking History, 5.2 (Summer 2001): 205). The Marwickian translation into the realist epistemological principle of 'know the archive, know the past, write the story' has been revealed as sham in theory and practice for a good many thinkers and history practitioners for some time now.

Ironically (but not surprisingly to me), in his desire to critique the auteur theory of history Professor Marwick has to spend rather a lot of time on the faults of individual historians (auteurs). As I have said already, Professor Marwick's anti-a priori and remarkably tenacious pro-empirical and analytical epistemology often takes the shape of pithy personal attacks on several individual historians who had tried and, in his judgement, failed to write sensibly (sense-ably?) about history thinking and practice (by the late 1960s). Notable among these were R.G. Collingwood and his The Idea of History (Professor Marwick's judgement is that the book was 'said to be brilliant, this being proved by the fact that you couldn't always follow what he was saying', xi), E.H. Carr and his What is History? ('a book which sold almost entirely on the basis of its [misleading] title alone' ibid), Pieter Geyl's Napoleon For and Against ('nave drivel' ibid), and W.H. Walsh's Introduction to the Philosophy of History ('easy to understand but had nothing on the basic activities of the historian' ibid). Professor Marwick is more benign with Marc Bloch's The Historian's Craft ('brilliant' but tragically unfinished and tarnished by 'an incomplete and misleading translation' ibid) and Geoffrey Elton (the closest we get to a hero when Marwick writes in the form of a romantic emplotment?) was only just starting 'his magnificent exposures of the nonsense spouted by philosophers (including Collingwood) and Marxists, particularly Carr' (ibid). Having said that Elton does not escape a Marwick wigging because of his emphasis on the primacy of political at the expense of social history. As rapidly becomes clear, this initial list of historian's to be named and shamed far from exhausts Professor Marwick's anguish and angst at the failures of most historians to do the job right (i.e., the Marwickian way).

Marwick says that he never wants the last word and that he is always eager to have weaknesses in his arguments pointed out (xii). So, I would suggest that he misconstrues the nature of the constructionist self-consciousness about the process of structuring historical knowledge held by many historians in the past thirty years or so. Not only does he dislike the idea of being a reconstructionist, he is particularly upset about the use of the verb 'to construct' (10). His reconstructionist anti-constructionist (anti-social theory/philosophy) position is pretty plain. Any 'overuse of the phrase [to construct] simply evacuates from it of all significance, or operates as a barrier to teasing out in detail what is really happening' (ibid. my italics). The notion that the data can give us the meaning (presumably because we stand outside its realm to observe it - the spectator theory of history) of 'what is really happening' has rarely seemed more unconvincing than in this blustering and abusive book.

Marwick's effort to circumvent any sense of construction in history (a position forced on him by his authoritarian empiricist and reconstructionist theory of knowledge) compels him again and again to name and shame a particular group of constructionists for whom he has particular disgust - Marxists. For example, his denial that gender is 'culturally constructed' is because for Marxists (the arch-constructionists) it is the 'usual suspect' the 'bourgeois power structure' (ibid.). Apart from the debate about whether there is ample empirical source material upon which to construct such an argument and that history is theory all the way through (Note: I recommend after Kant and Bevir Marwick reads Alex Callinicos, arguably the most able Marxist theoretician working today especially if he intends to reply to this review) Marwick stridently continues to insist the data can be accessed without any prior theoretical input on the part of the historian (sorry, auteur). Frankly, the notion that Marwickian access to past reality has somehow pulled the rug from under the Marxists (and all constructionists of whatever persuasion, i.e. most historians working today) remains implausible.

Professor Marwick's fear and loathing of philosophy actually precludes him from trying to support his source-based history with the thinking of those realist philosophers of history who might help him. After Kant, Bevir and Callinicos Professor Marwick would next benefit from reading the likes of Martin Bunzl, C. Behan McCullagh, Chris Lorenz and Perez Zagorin (among others). While they are also wrong (from my perspective) at least they have more reasoned arguments than Marwick. If we do owe our historical knowledge to the models, a priori thinking and narrative making of historians (as I think we do) then Marwick needs to buttress his presently under theorised and weak arguments in a rather more sophisticated fashion than he does. He should at least begin with a consideration of realism rather than what he presently offers which is, it seems to me, is a mixture of anti-philosophy prejudice and misconception about the complexities of writing history.

Marwick's failure to engage with philosophy and much of the debate over the nature of history (realist and post-realist) results too often into his descent to pretty low levels of misrepresentation. For example, at one point in his glossary (which might be a rather jolly Devil's Dictionary if it were done tongue-in-cheek) his 'definition' of postmodernism holds that it derives its view of the world from 'Marxism and maintains that the societies we live in, being bourgeois, are evil' (293). Here we have a case of prejudice rather than serious thought. Such examples of his fundamentalist epistemological as well as political narrow-mindedness and simple intellectual confusion reveal much about Professor Marwick's misunderstanding, misconstruction and misapprehension of the complexities of doing history.

There is a common saying that when you are in a hole you should at some juncture cease excavating further. Marwick's failure to adopt this policy is embarrassingly revealed in some of his throw away sentiments. I am astounded that he should say, for example, that John Warren is a defender of postmodernism, or that postmodernists don't write books, or that historians are not idealist or materialist, just historians (on p. 41 he starts to use a JCB rather than a shovel) or his truly bizarre notion that 20% of a historian's work is likely to be the product of personal prejudice. By this token, Marwick's logic seems to be that, as he says, if you read four or five books on the same subject you will get a 'pretty objective account' of that subject. An equally implausible thought (from my perspective) is his association of history with the natural sciences (history is 'a scientific activity') rather than literature (249). There are many other regrettable /ludicrous/bizarre/funny/plainly wrong statements of this kind. I would like to say the book is worth the sale price just to find them as, dependent on your point of view, you may find them diverting, but as I got the book free to review I can't really comment.


So, for Professor Marwick the sources are elemental to generating truthful historical knowledge. Of course they are important. I don't know of any historians (Marxist, postmodern, Foucauldian, or whatever) who endorse the metaphysical relativism Marwick is rejecting (although he doesn't use the term). Metaphysical relativism doubts a reality independent of our minds. It holds our concepts (and/or language?) create reality for us. All things and objects (like the past) are, therefore, mind and linguistic creations. It means reality is that which exists only for the individual or that there is nothing outside the text (a position usually associated with Derrida but is something of a bastardisation of what he meant). It is this that allows the denial of the real. To repeat, I don't know of any historian who is a metaphysical relativist (including me).

Epistemic relativism, on the other hand, views knowledge of the real as being cultivated by our ideas and concepts. In addition, epistemic relativists recognise such ideas and concepts are unavoidably influenced by linguistic, spatial, cultural and ideological compulsions. As an epistemic relativist, my recognition of the margins of truth does not, therefore, make me the puppet of either prejudice or incoherence. In acknowledging the sources and the weight of language as an ideologically drenched discourse, epistemic relativists (like me) do not deny there is no proximate truth in history. But there is more than one way to get at it and, for that matter, represent it.

Marwick's evangelical empiricism is the result, as I have argued, of his prior rejection of a priorism and 'theorising' (xiii). Clearly, Marwick must (incoherently?) endorse a hard-core realist variant of the theory of empiricism in order to reject all other theories, which are, by definition, antithetical to his preferred view of the (80%?) knowability of reality? Professor Marwick's reluctance to recognise empiricism as one among several available theories of knowledge is, of course, instrumental in producing his narrow view of how we can come to terms with the past as history.

In his defence of sources Marwick is forced into the difficult position of trying to have his factual cake and eat it. For some unfathomable reason he dives into some very murky waters as he feels obliged to deny historians discover 'the facts' saying (correctly I think) they are not atoms of past reality. This is an indisputable position for a historian to take (and is something even analytical philosophers have been saying for some time now) and its recognition (of what is so plainly obvious?) makes a welcome development in the latest edition of Marwick's book. He did talk about facts thirty years ago implying there was such a thing as a corpus of them from which the historian could select and organise their interpretation (1970: 143). But the epistemological sentiment that gave rise to that (odd even in 1970) view remains in the new text.

I am not entirely clear now what Marwick's attitude is toward facts. In saying they don't exist (generally) he still describes 'simple public facts' (the date of the Battle of Hastings') and 'complex private facts' (psychological states) (152). So there are things called facts? He then asks where the facts 'come from'? His answer is that they 'come from' the 'traces' left by past societies (153). They 'come from' the 'primary sources'. In the end, presumably to avoid the knotty problem of what is the epistemological status of the sources and the facts that 'come from' them, he falls back on his key words 'knowledge' and 'evidence' with the former based almost exclusively, of course, on the latter. Essentially, I think he is saying the facts (and he continues to insist we should abandon the term even though he gives examples of at least two kinds) arise from the interconnections worked out between events (the structure as embedded in the evidence) by the historian. At the end of the Marwickian process of forgetting the facts and foregounding the sources (Chapter 5), historical knowledge 'has to be garnered' indirectly by inference. At this stage in his thinking I think he is saying facts are interpretations and, despite myself, I'm suddenly feeling more pre-disposed toward Marwick. But I have an uneasy feeling even as I read that, that because he is an empiricist recidivist, he will disappoint me in later pages.

From my epistemically relativist standpoint facts (I'll say there are facts and they are our inferences which we cast as short stories within larger narratives) are a part of the empirical-analytical-narrative-linguistic-cultural process of doing history. As the philosophers Frank Ankersmit and Arthur Danto have argued, facts are best thought of as events under a description or, if you prefer, mini-interpretative narratives. As Peter Munz argued (not to mention Paul Ricoeur) to do justice to time (the essence of history is the representation of change over time?) it has to be described in the form of a narrative or a series of mini-narratives (Munz 'The Historical Narrative' in Michael Bentley (ed.) Companion to Historiography, 1997: 851-872). For Munz (like Ricoeur) the narrative form is the only way we can approximate past time's structure (852). This suggests facts remain the building blocks of history but defined rather as small narratives about events and happenings making up the larger emplotted historical narrative that constitutes refined historical knowledge.

The sources are the starting place for doing history (otherwise it is not history?) but the sophisticated historian quickly cranks up to the level of narrative. Marwick struggles here. He wants to accept that facts are not bits of reality to be discovered (why does he spend so much time denying what is patently such a silly idea?), and yet he cannot bring himself to accept that they are elements in a bigger fictive narrative (he prefers the notion that narrative is just a chronology of events and so the historian has virtually no input into its construction - sorry, structure). Facts are clearly the synthetic constructions of historians. All events (the election of Lincoln, the publication of Marwick's book, the market price of wheat in 1815 in England) are all elements in the larger narratives of various tragedies or romances? These mini-narratives are capable of comparison and verification. They are descriptions of things that according to the available evidence historians think happened. But what they mean is ultimately the product of the larger narrative of meaning as invented and described by the individual (auteur?) historian.

In the view of Hayden White facts are always re-shaped by the historical narrative. E.H. Carr viewed facts as not speaking for themselves but having a voice only through the historian. I agree with Carr when he said the historians' work is not just about colligation, the act of getting the story straight. As soon as the historian selects evidence the Marwickian notion of objectivity (getting the story straight) is compromised. There is no objective quality to the process of evidence selection. There is no invisible yet empirical hand that directs the historian as Marwick argues. Writing a historical narrative is a creative process. It results from the historian's imagination as much as it is the garnering of any structure or meaning that he or she may or may not believe already exists in the source.

Being an epistemic relativist I do not doubt the existence of the sources. By the same token I doubt we can or, what is worse, must only rely upon Marwick's 'consistent professional standards against which all history can be measured' (293). The narrative-linguistic additional dimension to history thinking and its procedures and principles in addressing the sources that I advocate means the facts, figures, numbers and record must act as a check on the history we write, but it does not wholly decide the form of our history. The narrative-linguistic answer to whether there is only one true story in the sources is that the narrative provides a meaning based on the manner in which the historian has imaginatively prefigured the historical field and, thereby, selected, composed, organised and interpreted the data. History results from that authorial and intentional act. History is not something like a script from the past which contains its own meaning or 'truthful interpretation'.


As I hope is becoming clear, I believe we can't divorce the three elements of doing history that Marwick isolates. If we assume, as I do, that history is at its most basic a narrative about the past rather than any privileged reflection of it through its traces, then Professor Marwick's denial that historians use language in the same way as writers of fiction, novelists, dramatists and poets and yet that history is not a reconstructive representation of the past becomes of key importance (12). According to Marwick historians should and can convey their meanings unambiguously. Generally he insists metaphors ought to be avoided. This is a position on which he is unchanged since 1970. It is a position akin to Voltaire's who also disliked metaphors and who said 'too many metaphors are hurtful...to truth, by saying more or less than the thing itself' [quoted in White's book Metahistory: The Historical Imagination in Nineteenth Century Europe: 53]). The epistemology that gives rise to this odd idea about language as a complex but controllable and decipherable mode of communication, also demands there is a match possible between the past world and the present word assuming we write clearly without too many metaphors (the correspondence theory of truth/knowledge). Of course Marwick's explanation and understanding of the nature of metaphor is incomplete. He seems to think in terms of figures of speech (metaphor, synecdoche, metonymy, etc) rather than what they represent epistemologically which is figures of thought (technically the theory of troping or root-metaphor). After Kant, Bevir, Callinicos and the rest I suggest he also reads Stephen Pepper's World Hypotheses (1942) as the seminal work on the connections between language, history and truth. He may also wish to consult Paul Veyne's Writing History, Essays on Epistemology (1984). Taken together Pepper and Veyne argue that all inferences, arguments and narratives are located in a root-metaphor that holds within itself a theory of truth, reference, perception, methodology, etc and which is the primary mechanism for coping with experience (past and present).

Hence, Marwick fails to explore (at all) history's fundamental narrative-linguistic character by considering the historian's capacity to think and corroborate evidence (empirical data) in language. He does not discuss or even acknowledge the role of the historian's consciousness, expressed mentally as 'figures of thought' as the means through which historical explanations are generated. Such ways of thinking (metaphoric, synecdochic, metonymic, etc) I believe directly influence the history making enterprise as a whole including its referentiality (choice of sources) and the theories, conceptualisations and arguments the historian deploys. In response to Marwick's three elements I think there are four key aspects of historical thinking and practice that need to be addressed, and which are essential to an understanding of how historians come to terms with the changing past. They are the epistemological foundations chosen for historical knowing; the role of referentiality (the 'reality of empiricism') in constituting the discourse of history; the deployment of theory and concept in creating an understanding of what the past means; and the figurative processes of writing and representation that constitute meaning (Munslow, Deconstructing History 1997; Munslow, The Routledge Companion to Historical Studies, 2000).

Hence, it is my judgement that the narrative connections made between representation and interpretation and, ultimately, the creation and assertion of meanings is as important to doing history as are the sources upon which the Marwickian idea of history rests, that is, the idea of the past as given in the sources. My position is that history is not simply an observational and reconstructive empirical activity. It is also a mind- and discourse-dependent intellectual and literary act. History is not the description of a given indubitable past. Certainly we can choose to believe the past once existed (I choose to believe it did). We can also choose, as Marwick does, that it can be 80% knowable. I choose to disagree with that on epistemological grounds.

Anyway, pretty close to true knowledge is feasible, so Marwick in his dogmatic empiricism believes, if the procedures and principles historians use in dealing with their sources are combined with the fundamentals of good writing (reduce the metaphors) (Chapter 6: 195-240). Language and narrative form are always secondary to the archive for Marwick. Because his argument is that the historian's narrative exists only to get the chronology correct this means it cannot serve any other useful epistemological purpose. For the ideal Marwick historian after the detail of their chronological narrative of 'this happened and then that' comes the description of events and action and, finally, the inference of what it all means (analysis). His assumption is that narrative, description and analysis are wholly separate epistemological activities. This is where he is, as I have suggested, seriously mistaken. Although he describes A.J.P. Taylor as 'ridiculous' (15) I believe Taylor was moving in the right direction in arguing that history is a form of story telling. I would go further, however, history is a literary genre. If I am right, this means historians must become more aware of how they constitute the stories they believe the sources are likely to be telling them?

Another of Professor Marwick's many misconceptions is that the narrative-linguistic approach to doing history is intended to somehow replace proper history with a delinquent or degenerate form based on the belief that there is no past reality, only source-less and, therefore, wholly fabricated historical narratives. This is not what is being argued. True, this can be done if the purpose is to generate a sense of the past and referentiality is deliberately made more tenuous. This is, I think, a legitimate extension of the boundaries. But the point that White (and many others) is making is surely fairly clear if thought about with a little care. The data or sources of the past, which it can be argued are already constructed narratives, are taken by historians and re-described. Historical knowledge is literally crafted in the etymological sense, as White says, of 'fictio'. The thing to realise is that the narrativisation of the past is a fictionalisation 'insofar', according to White, 'as narrativisation imposes upon reality the form and substance of the kind of meaning met with only in stories' (Hayden White 'An Old Question Raised Again: Is Historiography Art or Science?' Rethinking History: The Journal of Theory and Practice, 4:3 (2000): 399).

White's insistence on the cognitive parity of form with content is not, of course, original. Benedetto Croce and R.G. Collingwood both insisted that what historian's imagine provides the connection between past reality and its representation as history. Both Croce and Collingwood understood that the historian's principal mechanism for thinking about the past and its content is the historical imagination. As Collingwood said metaphor (as a particular figure of thought translated into a figure of speech) allows the historian to produce a mental picture of things past. Metaphor is the interpretative link between the past and the historical narrative. Rather than being an obstacle to historical thinking, as Marwick believes, metaphor is the essential feature of all thinking including the historical. I believe that, apart from Marwick, most historians would agree that knowing the sources and comparing and contrasting them, verifying and relating them does not equate with knowledge of their intrinsic meaning. Rather than discovered or recovered, 'garnered' or 'come from' it is the historian who provides the meaning of the past in that narrative we call history.

The narrative-linguistic contribution (of Hayden White among many others), therefore, is the recognition that, as history is a discourse about the past-thing-in-itself, then it is the historian who must first imagine and conceptualise it before they can bring to bear upon it the Marwickian methods and principles they want to employ to explain it (if that is what they elect to do). As White argues, the historical narrative is characterised by a double representation, that of the past and the historians thinking (figures of thought) about how to compose it as a narrative, literally the metanarrative. Hence, making sense of the past cannot be done by reference to sources alone for there is an unavoidable poetic or story-making element in writing history.

Most serious thinking historians these days realise that narrating the sources means history can never replicate the past. For empiricists like Marwick this is antipathetic. But surely history, whether a history book, lecture or article, or a book review like this, is some kind of a representation? What is meant by representation? Marwick's deep confusion is evidenced at the end of chapter six on the communication of historical knowledge. In an attack on (my friend and co-editor of the journal Rethinking History) the leading historian of film Robert A Rosenstone, Marwick says it is absurd to say as Rosenstone does, that a film can provide a construction of the past. Instead Marwick much prefers the view of Mark C. Carnes that like drama and fiction movies entertain and can give insights into the human condition. But they do not provide a substitute for history 'painstakingly assembled' from the best available evidence and analysis (quoted in Marwick p. 239). This is precisely the point Marwick (and Carnes his 'source') does not grasp. There is, epistemologically, nothing peculiar to history that gives it superiority over any other form of representation. All representations of the past are constructions (painstakingly assembled?).

Surely history does not resemble or mimic the past (defined by Marwick as what actually happened and is now 'gone for good' (292)); it is a substitution for it? As such, a history becomes, ontologically speaking, a linguistic thing in itself (like film it certainly and obviously isn't the past or, for that matter, a resemblance of it). My logic here is that (speaking epistemology) referentiality is that which connects words to things, and representation is that which connects things (like the past) to things (like history). History defined as a historical representation (a linguistic thing/entity) has no reference to the past as such; at best it is about the past, it is a linguistic substitution for it (like film is a celluloid substitution). That which is now gone is replaced by the historian's composed historical narrative. History, thereby, substitutes for the absent past. Once we realise this we have an additional level of armour against those empiricists who would lie (Holocaust deniers readily come to mind) through a selective mangling of the sources.

Marwick will, of course, continue to insist that only after we examine the sources can we justify our descriptions in a narrative-linguistic form. But, my response is that whether we like it or not, the fact that the past has gone and now has to be narrativised as another thing, a something else that is a linguistic entity (that we choose to call history) means we are operating according to a substitution theory of representation. Ultimately, the givenness of the past is something imputed in the brain of the historian. To argue, as Marwick does that all historical knowledge must start and end with the sources implies that to doubt their epistemological functioning means we have nothing left upon which to do history. This argument is akin to saying we must accept photocopied 5 notes for fear we might not have any money. There is no danger that we shall ever be out of touch with the past because we have its substitute: history. The problem is that history is a descriptive substitution for it.

My objection to Marwick's version of procedure-based empiricism occasions me to ask why is it that he invests so much effort in trying to deny that it is both the past and its substitution as history (a linguistic entity) that together generate historical knowledge? To claim, as Marwick does, that the turn toward the narrative-linguistic is useless (or worse), is to perpetuate the error of conflating knowledge of what the sources mean with their description. I fear Marwick's book will extend the life of the highly dubious argument that the journey from the sources of the past to the history text does not affect history's truthfulness or objectivity (well, up to 80% of it?).


I suppose it could be thought that any book that prompts me to write something in excess of seven thousand words about it must be well worth reading? Well no, I don't think so. What has justified my effort through a good part of late July and early August 2001 in (reading the book and) writing this review is my belief that Marwick's position requires some sort of reply. I shall not comment further on the mean-spirited and condescending tone with which it is written. What is important is that Marwick has offered a defence of source based historical study that accepts the existence of data as the ultimate touchstone for meaning. The implication is that proper historical research allows us to know the reality of the past as it once was. Despite the most recent efforts at recouping realism (see C. Behan McCullagh and Mark Bevir) it seems improbable to me that we can, as historians whether as a profession collectively or as individual auteurs, know the past as well as people in the past knew it (even given the 80% test) even though we concentrate on the sources and their corroboration. This is the epistemic relativism that forms the basis of my kind of history.

If I am right, and we cannot know the past independent of our conceptualisation and narrative composition of it, the possibilities of 80% accuracy (which would be pretty damn good) are slender. No historical account can tell us what the past was. All we have is a narrative about it. Like Marwick historians can elect to be sceptical empiricists, i.e., they don't take the sources at face value but who believe they can more or less reconstruct the past as it most like was and hence garners its most likely true (oxymoron?) meaning. Or we can choose to be more realistic by confronting the improbable nature of the discipline. The fact that the vast majority of historians are constructionists of some sort or another (to choose a percentage how about 80%?) demonstrates that sceptical empiricism is not enough. The only history we can have, because we produce it, can never satisfy the correspondence theory of truth. All that historians can be happy with, I think, is a narrative truth. To argue otherwise, as Marwick does in this book is, paradoxically, to be unrealistic and misleading. To suggest that empirical data is self-justifying because it can be understood without a theoretical or narrative-linguistic dimension makes little or no sense to me. An epistemological position of doubt, not disbelief as Marwick implies, about the value of the empirical in doing history is, for me, the most realistic appraisal of its nature.

September 2001


1. A point made to me by Peter Munz in a letter dated June 12th 1999.

Author's response

Back to reviews index

Back to top