Nothing could thus be clearer than that the narrativists reject the picture or copy theory of representation. The bad news is that we stand to lose a great deal in terms of historiographic epistemology. Namely, if the past was like the photographer's object waiting there to be immortalized in a representation, one could speak of discovering it and capturing it as it (really) is or was.6 All that one needed to do in that case is ensure the accuracy of our depiction.
Unfortunately there is no 'discovering' the past. A more appropriate metaphor here is that of 'construction', as becomes clear in the following:
The 'historical landscape' is not given to the historian; he has to construct it. The narration is not the projection of a historical landscape or of some historical machinery, the past is only constituted in the narratio. The structure of the narratio is a structure l ent to or pressed on the past and not the reflection of a kindred structure objectively present in the past itself. (Ankersmit 1983, 81)
There is more than one possible reading available concerning the sense of saying that the 'historical landscape' is not given and requires construction: sociological, epistemological, and metaphysical.
The first option deals with actual historiographical practice. The past as a 'historical landscape' is not available to the historian, as mountains and seashores are for the cartographer. One cannot compare and model one's representation to any tangible and observable object. For this reason, the historian has no other option but to construct a narratio in the most concrete terms. Historiography is thus in a trivial sense a constructivist endeavor as opposed to research that discovers or finds that which exists there prior to any investigation. This is the sense in which the historian constructs a narratio, which is 'lent' or 'pressed' onto the past, and we might say that only then does the past become intelligible.
I call this first reading of constructivism sociological in the sense that it relies on an (arguably correct) observation of the constructivist nature of historiographical practice. If we were to send a sociological observer, the kind of figure in Latour and Woolgar's (1979) Laboratory Life: The Construction of Scientific Facts, to observe how historians work, the observer would certainly not find that they copy and compare their representations to some pre-modeled past in front of them. Indeed, it is safe to assume that not even the most committed narrative realist would object to sociological constructivism. Realists are prone to common sense views and it would just be too fanciful to suggest that the ready- structured past somehow floats into view in front of the historian's eyes, in order that it can then be mimicked in a representation.
However, realists are likely to try to turn this state of affairs to their advantage. An example of how this could be done is Geoffrey Elton, a dominant figure together with E. H. Carr for decades in seminars on the theory and methodology of history and historiography. Elton's point is that the historian's research object, the past, is strictly independent of the historian's inquiry and therefore unmodifiable. In this sense, historiography is in a better position than many natural sciences! For example, in biomedical sciences scientists interfere with their research object and thereby jeopardize its independence of them. Elton writes, 'Just because historical matter is in the past, is gone, irrecoverable and unrepeatable, its objective reality is guaranteed: it is beyond being altered for any purpose whatsoever' (1967, 53-54). Practical difficulties in reconstructing the past may boil down to the uncertainty and the insufficiency of evidence in establishing facts. For Elton there is always truth to be discovered although not necessarily always to be found in practice.
This discussion has brought us into the domain of historiographic epistemology: how can we acquire historical knowledge? Given that the past is not accessible in any direct and unmediated way, perhaps one could nevertheless reconstruct it and its meaning with the help of traces left from the past? Indeed, Elton and historical realists imply that we can bridge the gap between historians and the past they investigate. Further, we might say that much theoretical discussion in modern historiography has focused on formulating methodological advice on how to deal with source material, that is, what kind of source criticism would guarantee epistemically warranted - and even beyond this, true - conclusions. Elton writes that the 'Historical method is no more than a recognized and tested way of extracting from what the past has left the true facts and events of that past' (1967, 65). And it is worth remembering Ranke's injunction that historians should extinguish themselves, in order to prevent their subjective beliefs and assumptions from being projected onto their reconstructions of the past. It is much better on this view to let the facts found in documents speak for themselves.7
The narrativists represent the opposite end to modernist historiographers, at least when it comes to narratives and other synthesizing interpretations. Using Ankersmit's terminology, we might say that a narrative realist, like Elton, believes that there is something akin to 'translation rules'8 that govern the relationship between the past as given and the representation of the past as represented by the historian. Most admit that there are no direct rules of correspondence, but many insist that indirect rules, such as appropriate reading of source materials, exist. By contrast, White abandoned the narrative realists' attempt to understand the past itself and their belief that there are such translation rules that show what historical reality really is like. In White's theory, his famous four tropes (Metaphor, Metonymy, Synecdoche and Irony) assume the role of translation rules, which are culturally embedded in our way of making sense of the past. However, they do not reveal the true nature and shape of historical reality, but function like a Kantian 'transcendental deduction'; in this sense, the tropes are needed to make historical knowledge possible in the first place. Ankersmit commits to the same idea of the transcendental limit of historical knowing. The difference is that he is clearer that the translation rules in this sense are not proper translation rules of the past at all. They reveal to us, not what the past is like, but only show the logical structure of the narrative accounts of the past (Ankersmit 1983, 77). Ankersmit writes that 'Whatever concrete content we may give to the translation rules, they will never be more than arbitrary selection rules, acceptable to some historians but to be rejected by others' (1983, 81; my emphasis). Or simply, 'there are no translation rules' (1983, 87; similarly 216, 226).
The statement that 'there are no translation rules' is a clear denial in the context of epistemology that one could have (even) indirect epis- temic access to the past, which would show it wie es eigentlich gewesen. However, the debate between the realist and the idealist does not stop on the epistemic level either, which is thus the second sense in which the 'historical landscape' is not given and has to be constructed by the historian. The quotation above (p. 37) continues as follows:
The past is by no means like a machine: it does not possess some hidden mechanism whose working the historian has to trace. Nor is the past like a landscape that has to be projected onto the linguistic level with the help of projection or translation rules. . .. We should reject "the idea that there is a determinate historical actuality, the complex referent of all our narratives of 'what actually happened', the untold story to which narrative histories approximate" (Mink). ... All this means that the past as such has no narrative structure - narrative structures occur only in the narratio. (Ankersmit 1983, 81)
Because there is no narrative structure or any other 'untold story' in the past, there is nothing to tell and nothing to discover, even if we had the 'access'. The past only becomes narratively structured through the imagination and the hand of the historian, who imposes order and meaning there.
It is notable that both White and Ankersmit assume that the individuation of historical facts is unproblematic, an example of which is a chronicle. As the latter expresses it, the correspondence between individual statements of a narratio and historical reality is 'beyond doubt'. Ankersmit writes that 'Saying true things about the past is easy [on the level of individual statements] - anybody can do that' (Ankersmit 1990, 278).9 But everything is different with historical texts. There is no story 'within the welter of facts' (White 1975, 59). 'The sets of relationships' that the historians postulate are not 'immanent in the events themselves; they exist only in the mind of the historian reflecting on them' (White 1978, 94). Yet more pointedly, no historical event is intrinsically tragic, comic, and so on (cf. White 1978, 84). We can understand White here, first, making a negative point about historical reality. Whatever the past is, it does not inherently possess any narrative or story forms in which historians present their accounts. This is a comparable position to Ankersmit's narrative idealist, who thinks that 'the past as such has no narrative structure'.
But there are some indications that both go yet further than this and maintain that the past has no structure whatsoever. The most obvious point of reference is the theoretical chapters in Metahistory, where White writes that the historian has to prefigure the field of investigation before it can be a possible object of knowledge. Prior to that constitutive act, neither objects nor the relationships between them are constituted. This also is the reading of White that Ankersmit builds on in his writings. According to Ankersmit, White understood the past as 'a meaningless myriad of facts, states and events, an amorphous chaos of data' (1983, 78), which is an idea that re-appears in Ankersmit's later books. He writes that historical reality remains a chaos as long as a representation has been singled out to bring order into this chaos (2001, 45) and that the historians' concepts create continuity and unity in the field that is in the state of chaos and disorder before their postulation (2012, 45).
The stronger interpretation of their position is thus that White and Ankersmit not only state that the past is not narratively structured but also deny historical reality any inherent order or structure whatsoever. Ankersmit's words, above, that there is no 'determinate historical actuality' at all might be read in this light. This is to claim that they take a metaphysical stand on the nature of reality, which could perhaps be described as nominalism.1 0 Perhaps White and Ankersmit should be understood as advocating the unorthodox form of nominalism, something like that which Ian Hacking brings forward in his The Social Construction of What? Hacking uses the term 'nominalism' to individuate a position that rejects, above all, that nature has any given natural kind structures. Hacking calls the opposite view 'inherent-structurism', that is, the belief that the world comes with an inherent structure (Hacking 2001, 82-84).11 The writings of both undoubtedly contain ingredients for this kind of interpretation, but one should note that the idea of the past as chaos is related to the historian's practice. This suggests a slightly different reading, which deals not with the nature of historical reality as such but with the historian's role in making sense of this reality, saying that whatever the true nature of reality is, for us, and for historians, it appears chaotic before we order it with our concepts and narratives. In other words, Ankersmit and White may be talking about the 'chaotic' order of historiographical data and not about the past as such.
The claim that narratives are not found in the past but imposed on it reveals yet another fundamental aspect of narratives in the narrativist philosophy of historiography. White has famously stated that historical narratives are 'verbal fictions, the contents of which are as much invented as found' (1978, 82) and that the historical work is 'a verbal structure in the form of a narrative prose discourse' (White, 1973b, 2). White used these remarks to justify his comparison of historiography to literature. Further, White also writes that the qualities of narratives are only something that people seek and value, and that they do not have any correspondence in the past:
This value attached to narrativity in the representation of real events arises out of a desire to have real events display the coherence, integrity, fullness, and closure of an image of life that is and can only be imaginary. The notion that sequences of real events possess the formal attributes of the stories we tell about imaginary events could only have its origin in wishes, daydreams, reveries. Does the world really present itself to perception in the form of well-made stories with central subjects, proper beginning, middles, and ends, and a coherence that permits us to see 'the end' in every beginning? (White 1987, 24)
Here emerges the position that Ankersmit has expressed philosophically in less ambivalent terms: 'Given this morphological or structural difference between the past and the narratio, how can translation rules ever be expected to link them together? Projection or translation rules can exist only where there are two corresponding spheres of structural similarity' (1983, 82; my emphasis). Ankersmit refers in this context to Poincare, who said that we can compare one clock to another, but not to 'time' itself, because there is no time as such. The same reasoning is said to apply to historical narratives. 'We cannot glimpse at history. We can only compare one book with another book' (1983, 81-82).12
In other words, there is a morphological or structural difference between the historian's presentation and historical reality, which explains why any idea of copying or matching between the two is fundamentally misconceived. One simply cannot make two structurally totally different entities correspond with each other. Elephants cannot be made to correspond with butterflies due to obvious structural differences. The historian's narrative is verbal and textual, while historical reality is nonnarrative and non-verbal in nature. We come to the conclusion that constructivism is not a forced option merely because of practical problems of copying the past, either sociological or epistemic, but due to a yet deeper, metaphysical, philosophical problem. White and Ankersmit wish to point out that whatever reality is like, the features of historical representations are not part of it. This is what Ankersmit's and White's constructivism fundamentally boils down to.
The gap between us and the past world is bridged by the historian's representations, which unfortunately results in the morphological divergence between the representation and the represented. This takes us directly to a common theme underlying White's and Ankersmit's philosophy, already briefly mentioned earlier in this book: Kantianism.
In the last chapter we saw how Ankersmit understands both philosophy of language in general and White's and his own early scholarly works as subprojects in the larger Kantian program. All philosophies of language allegedly have in common the idea that language is the principal condition for the possibility of all knowledge and meaningful thinking (1994, 2; similarly 2008b, 84). In White, tropes function like Kantian categories of understanding, as preconditions of meaningful historical knowledge.13 And in the early Ankersmit, this view comes down to a conviction that it is the metaphor that organizes historical knowledge and makes the unfamiliar familiar.
More specifically, in Narrative Logic Ankersmit made it clear that his approach in that book resembles Kantian philosophy in that it aspires to answer the question of how narrative knowledge of historical reality is possible. He says that it is an attempt to develop 'a Critique of Historical Reason', an obvious reference to Kant's Critique of Pure Reason.14 And although Ankersmit indicated that his intention was to move beyond Kantianism in History and Tropology (see chapter 2), he remains Kantian also in his later publications. In Meaning, Truth, and Reference in Historical Representation, Ankersmit again tries to resist the Kantian discourse, but succumbs to it under its appeal and pragmatic worth: 'I would rather avoid transcendentalist vocabulary ... this nevertheless is to a large extent the view inspiring the remainder of the book' (2012, 47).
Further, Ankersmit writes that historians, in their belief of finding correspondence between the past and historical language, are blinded 'to the fact that the unity and continuity supplied by historical language is the transcendental condition for the possibility of historical knowledge' (Ankersmit 2012, 46; my emphasis). Their poor vision might lead some, such as the French Annales historians, to prefer an 'incoherent raw mass of information' to more developed and succinct accounts (2012, 46). But it is the historical representation as a Kantian transcendental condition that in Ankersmit's view brings order into chaos: outside historical representation there is no historical writing (2012, 46).
Finally, it is worth remembering how White, cited above, claimed that no historical event is intrinsically tragic. The story form encodes the facts of history into an intelligible narrative. Consistency, coherence, and 'illuminative power' as qualities of the historical account stem from the historian's vision and choice of the mode of presentation (1973b, 4). The same set of story elements that looks tragic from one point of view (through one story form) may be comic from another. Michelet construed the French Revolution as a drama of Romantic transcendence, but Tocqueville emplotted the same phenomenon as an ironic tragedy. White even calls the potential elements of story value neutral; meaningless events in serial order that are transformed through plot structures into meaningful structured stories (White 1978, 84-85; 1987, 44; 1975, 59). His words have a familiar Kantian ring: 'The implication is that historians constitute their subjects as possible objects of narrative representation by the very language they use to describe them' (1978, 95).
The status of historical accounts as 'possible modes of historical representation or conceptualization does not depend upon the nature of the
"data" they used to support their generalizations' (1973b, 4). Indeed, White thinks that the mode of presentation is an a priori choice, not to be judged empirically at all. These modes are reconceptualizations that make historical narrative and knowledge possible, without which historical knowledge would remain in the state of an incoherent heap of data. Story forms or tropes thus add something to the past that does not exist independently there prior to the historian's creative act (cf. White 2011 395, 397). And, because there is in his view a limited set of story modes available to the historian, the historian imposes on historical reality not only one's own personal vision but also a (Western) culturally conditioned encoding more generally.