Walsh writes that there are two conditions that govern the choice of using a particular colligatory concept. First, the concept must be 'tailored to fit the facts rather than a straightjacket'. This comes down to a requirement that generally accepted facts or statements about historical facts must be seen as supporting a colligation. A small number of disagreements do not show, according to Walsh, that a colligatory concept is defective. What matters is the overall support for the interpretative framework involving the colligatory concept (Walsh 1974, 139). The second condition is the question of how a concept illuminates the facts, which is the issue of how its 'use makes the past real and intelligible to us' (Walsh 1974, 140; my emphasis). The idea is that, given that the 'facts cannot speak for themselves', the historian has to organize the past and make it understandable through some synthesizing interpretations and concepts. Remember that earlier Walsh talked about how a set of facts become intelligible in the light of or through colligatory concepts by constructing a 'significant' narrative of the events (e.g. Walsh 1958, 62). In brief, the historian has a double duty: do justice to evidence and to readers. Colligatory concepts must be chosen in such a way that neither of these aspects is forgotten. Presentations of history should not be mere lists of factual statements without any communicable coherence, nor vulgar abstractions without any factual support. I will elevate the idea of casting new light or illuminating historiographical data5 to the status of the first criterion:
1. Exemplification: The descriptive content of a colligatory expression has to exemplify the historical data it subsumes.
In other words, the historian may use a familiar set of data as material for the colligation and, in this way, give a new meaning to this part of the past. What did Ehrenburg's metaphor of the 'Thaw' communicate? The idea of a thaw and its contrast to freezing implies the thought of a warmer and happier period. Indeed, in the novel, Ehrenburg's 'Thaw' refers to people becoming happier and freer due to a change in the intellectual climate that enabled them to think and act more spontaneously without constraints and a need for pretending. This seems to transfer well to the history of the Soviet Union. The 'Thaw' translates to people gaining greater freedom to think and act as well as to the process of relaxation of control in various arenas of life in post-Stalinist society. The historian wishing to use the concept 'thaw' would naturally therefore refer to events and phenomena that exemplify this sense, as already discussed above.
The implication of this is that the descriptive content of a colligatory expression must appear as appropriate with respect to the historical data, which entails a minimal requirement of truth. The colligated statements describing historical data should be assumed to be true, unless they are colligatory statements themselves. On the other hand, what appears as appropriate is relative to what each feature is seen to represent in the historian's own time. If we understand the release of prisoners as a form of liberation, and the release really happened (truth-requirement), then the 'Thaw' may be said to describe well what took place in the Gulag in the Khrushchev years. And if we interpret the Soviet leaders' visits to the West as friendly gestures, and they actually made such visits insofar as we can tell, these can similarly be used to illustrate the 'Thaw' in Soviet history. To emphasize this point, the sense of a 'warming climate' is not inherent in the historical data, but the imposed sense must nevertheless be seen to illuminate that particular episode.
The idea of 'exemplification' can also be elucidated by an illustration from McCullagh (2008). Although I think he is wrong to claim that the historian could discover the 'French revolution',6 he provides a useful example. McCullagh writes that Lincoln's outlawing of slavery as well as the Congress' outlawing of some forms of racial segregation and granting the right to vote can be seen as a process of increasing the freedom for African-Americans. Would any contemporary historian dispute that? But this is nevertheless not an automatic and natural judgment. For example, we need to assume a certain sense of freedom and think, for example, that voting increases freedom. It is contentious whether voting in the Soviet Union increased the freedom of workers in the same, or indeed even in any, sense. The fittingness of the colligation thus depends on the analysis and judgement of the historical material. Given our cultural conceptions, this specific freedom hypothesis seems justified when applied to American history.
It is also important to recognize that there is always a vast amount of historical source material that could be selected and colligated: notebooks, archived public records, various types of printed material, public monuments, memories, material artefacts, etc. The historian can naturally select only a small number of all the potential data to examine and report, even at the best of times. As a consequence, it is almost always possible to find contradictory material for a given colligatory concept in any period investigated. For example, it was not the case that all unjustly sentenced prisoners were freed in the post-Stalin era. And at least some of Khrushchev's seemingly liberal acts with regard to publishing were performed for tactical reasons: to underline the difference to the 'criminal' era of Stalin while upholding and encouraging a sympathetic interpretation of the Soviet state and its leaders. Would it be appropriate to describe the internal climate as a 'thawing' if the seeming liberalization was just a tactic to support the new regime and keep a tight grip on the population? In 1957, the Communist party reminded historians that it had no intention of tolerating 'liberal interpretations' (Mazour 1958, 244). Kenez notes that 'the writing of history was [still] strictly supervised' to evaluate the Stalinist past correctly, that is, to examine the past critically but not too critically in order not to delegitimize the post-Stalinist government (1999, 190). It is worth noting that the symbol of Khrushchev's 'Thaw', Dudintsev's critical Not by Bread Alone, was denounced as anti-Soviet in 1956, that is, in the early years of Khrushchev's reign. Further, Soviet tanks rolled onto the streets of Budapest with chilling effect in 1956. Would these phenomena be enough to freeze the thaw in historiographical parlance? We might also consider an entirely different colligatory concept, that of 'scientific revolution'. It makes sense to claim that scientific thinking changed distinctively in the roughly two hundred years beginning from the mid-sixteenth-century, when supernatural explanations were rejected in favor of natural ones and reason and experimental method replaced faith and dogmatism in science. However, as many historians of science have pointed out, this period demonstrates much continuity too, in some fields more than in others, and particularly in the form of experimental practice and instrumental application. Does this mean that this concept, originally applied to rotating wheels and to the wheel of fortune, is inapplicable as a colligatory concept to describe scientific change in these years?7
The point is not that it should be possible to find only one correct colligatory concept, or that no concept is apt, but that the historian has to make an interpretative choice that is intelligible in light of historical evidence. It is as Thomas Kuhn wrote that 'there is no such thing as research without counterinstances' (1970, 79) and as Imre Lakatos provocatively put it, any theory 'at any stage of development, has unsolved problems and undigested anomalies. All theories, in this sense, are born refuted and die refuted' (1978, 5). Another way to express the point is to say that there is an infinite amount of potential historical data, but that there is no given, epistemically privileged set that a historian necessarily has to take into account. One might say that the selection of data is itself the historian's interpretative choice. Walsh proposed that some historical material could be allowed to contradict a colliga- tory concept, if that concept otherwise enjoys 'overall support'. This is correct, but it would be unwise for the historian to concentrate on highlighting contradictions or gaps any more than a scientist puts effort into highlighting known anomalies when proposing a new theory. The second criterion governing the application of colligatory concepts is therefore that the material highlighted should form as a coherent whole as possible, which thus entails that one should avoid contradicting the descriptive content of a colligatory concept.
2. Coherence: The material highlighted has to be chosen and constructed so that it forms a maximally coherent set.
It is important to realize that some colligations are more fitting and appropriate applications to historical data than others (without the expectation that they be absolutely correct). If some were not, it would be hard avoid the conclusion that all colligations are merely random figments of the imagination. The coherence condition requires that one try to maximize the 'fit'. That is, to show that there are inferential connections between the elements of data or even that they constitute a unified whole. This is related to another virtue: an attempt to colligate as much historical data under a concept as possible. Thus the third criterion of application is:
3. Comprehensiveness: The concept that applies to a larger amount of historical data than its rival on the assumed historical phenomenon is preferable.
If a historian suggests that it would better to call Khrushchev's period the 'Cold Snap' or the 'Big Chill,' the historian would be confronted with and be required to explain a vast amount contradicting historical material, despite being able to adduce some supporting evidence as well. These concepts would be able to colligate less than the 'Thaw' and there would be more contradicting material to make them unfit as colligatory concepts for this period. What if someone proposed that we call what is now known the 'Industrial Revolution' the 'Agricultural Revolution'? It would not be difficult to draw attention to the documentary traces of factories, industrial policies, products, workers, etc. in the historical record to question the applicability of this suggestion. This is not to say there may still not be better ways to account for this historical phenomenon or that there are no other potential 'analytical' ways, to borrow Marwick's term (2001, 4), by which to divide the period into some other segments.
Yet another standard that can be used to prioritize between colliga- tory expressions is the scope of application - not to evidence of a given historical phenomenon but to historical phenomena themselves. If a concept applies to, colligates, or makes intelligible a large area or amount of historical phenomena, making the events and objects appear as one coherent item, the colligatory expression has a large scope of application. A large scope in turn makes it a powerful colligatory concept and can be a reason to prefer it over other concepts.8 This gives us the fourth criterion.
4. Scope: Everything being equal, a colligatory concept with a larger scope of application to historical phenomena is preferable to one with a more limited scope.
This is a virtue that cannot be applied in isolation from the previous three, since a large scope cannot compensate for a colligatory concept if it only poorly exemplifies the historical evidence it is intended to highlight, for example. Furthermore, a large or narrow scope of application cannot be seen as an unambiguous reason for preferring a particular colligatory notion because that choice ultimately depends on the historiographical rationale of a given work. Nevertheless, colligatory expressions often have a very wide scope of application, which increases their appeal, as with the 'Enlightenment'. What is gained through a wide scope of application may, however, be lost in specificity since a very broad colligatory concept may explain some local phenomena quite poorly even when it makes a large part of the past comprehensible (cf. Tucker 2004, 148; 152). The fact that different cognitive values and virtues may be in tension with each other has been recognized in philosophy of science from the beginning by Thomas Kuhn and others.9 It is not possible to construct an algorithm to determine the steps to take to reach a uniquely correct justification; instead, there is always a trade-off between different epistemic values, depending on the aims of the historian.
Scope of application is related to the final criterion: originality. Although it is difficult to spell out how to judge some expressions as being more innovative and original than others, these attributes are undoubtedly virtues in historiography. While one should not expect to pin down any point-by-point criteria for innovativeness and originality, they have to do with the cohesiveness of the views on the past proferred. This judgment is also tied to our cultural norms and valuations in the framework of historical discourse. Given that it was already customary to talk about the 'Cold War', the concept of the 'Thaw' appeared intelligible in the existing framework, while it was also refreshing at the same time. The final, fifth rule is thus:
5. Originality: Everything being equal, a more innovative and original concept should be preferred to a more customary one.
As above with the fourth criterion, this fifth criterion cannot compensate for the problems with application to historical data itself. Chapter 10 continues the discussion on the significance of originality in historiography and its relation to objectivity.
These five criteria amount to a theory of justification, which outlines how one can judge and choose appropriate colligatory concepts. To repeat, these criteria do not yield us an algorithm that could be used to determine, with certainty, what to construct and accept and what not. Neither can they be used to identify uniquely correct concepts, but they do, nevertheless, amount to both empirical and extra-empirical criteria to be used in choosing and ordering colligatory concepts. Indeed, we are not dealing with absolutes but a ranking between more and less appropriate concepts. What is needed here is a change of discourse from truth-functional language to the vocabulary of comparativity and rational ordering. These five criteria are rules of thumb that can be used to compare and understand which colligatory expressions are fitting and justified in historiography in the absence of rules of correspondence between the historian's presentation and the past.