In fights between the factions of the victorious French revolutionaries in 1792–1799, the accusation of scheming to restore the monarchy was the most murderous claim that could be made. In power struggles over who would succeed Vladimir Lenin within the USSR (1924–1938), the indictment of plotting the restoration of capitalism was deemed most ruinous. This was the case in the infamous ‘Moscow trials’ of 1936–1938 where Kamenev, Zinov’ev, Bukharin and other rivals of Joseph Stalin were indicted for conspiring to dismember the USSR and to restore capitalism. The indictments were false, but after some 50 years the USSR was dismembered and capitalism was restored, as happened in France in 1814–1815 to the Bourbon monarchy.

Despite all the differences between the French (or using Marxist parlance ‘bourgeois’) and the Russian (‘socialist’) revolutions, they were productively compared in the comparative historical sociology of revolutions. Actually, comparison of these two revolutions has been seminal (cp. Sorokin 1925; Brinton 1965 (1938); Moore 1966; Paige 1975; Eisenstadt 1978; Tilly 1978; Skocpol 1979; Goldstone 2001) and remains paradigmatic in this academic enterprise. I believe that a comparison of anti-bourgeois and anti-socialist restorations can be just as productive for the extension of the sociology of revolutions by looking at the sociology of restorations.

Strangely, the work of the Austro-American legal scholar and historian Robert A. Kann (1906–1981) still remains the only attempt at an encompassing theorisation of social restorations (Kann 1961, 1968, 1972, 1974). Kann’s main research area was the history of the Habsburg monarchy, while his comparative historical study was most probably inspired by the political interest in the possibility of the restoration of the Habsburg empire, even without the Habsburgs, as a federation or confederation of Central European states. In his large study (Kann 1968), he compares 11 selected cases from Western history, starting with the restoration of the Jewish homeland after their release from Babylonian captivity by the Persian king Cyrus the Great in 538 BC and closing with the restoration of the German Empire in 1871. Table 2.1 provides a complete list of Kann’s cases.

Table 2.1 Robert A. Kann’s set of restorations. Question marks are inserted where Kann does not provide the chronology of the original or restored systems, and figures were inserted using data from standard reference works
Full size table

I take from Kann the idea of restoration as the final component in a larger pattern of social change, featuring the sequence of original (A), intermediate (B) and restored (C) social systems, where the restored system affirms, constructs or claims continuity with the original (or ancient) system that was disrupted by the revolutionary transition from the original to an intermediate system. I share with Kann also an interest in the causal conditions of restoration success. Indeed, Kann not only provided the first conceptualisation of social restoration but also formulated at least two causal hypotheses, defining restoration success as the durability of restored systems. Successful restoration regimes endure. Restoration regimes fail if new revolutions break out, although occasionally restoration regimes may be saved by foreign interventions. These are Kann’s ‘laws of social restorations’:

  1. 1.

    Success of restoration is positively related to the duration of the original system.

  2. 2.

    Success of restoration is negatively related to the duration of the intermediate system.

These hypotheses can be statistically tested using the dataset in Table 2.1. In one test, I tested correlations between duration (in years) of A and C and B and C. Alas, both correlations (Pearson’s r) are very weak. Besides, the correlation between the duration of B and C has the wrong sign (i.e. the correlation contradicted Kann’s hypothesis). In another test, logistic regression was applied, coding restorations as successful (1) or failed (0) according to Kann’s judgements, which in some cases seem to go against his own criterion of endurance of a restored system over more than one generation.

However, binary logistic regression misclassifies two out of four restoration failures and one out of seven restoration successes using duration of the original system as an independent variable, while the regression coefficient has the ‘wrong’ sign. Using duration of the intermediate system as an independent variable, the sign is correct, but regression misclassifies all failures as success cases, yet does not make mistakes for success cases. Most discouragingly, all the results of both tests are not statistically significant at the p = 0.05 or p = 0.1 level.

This could only be expected given the very small N size. Actually, Kann himself did understand his hypotheses as deterministic statements, describing the necessary conditions for successful restorations:

  1. 3.

    Restored system C cannot endure for at least two generations (70–80 years), which means its restoration is successful if original system A endured for less than one generation (35–40 years).

  2. 4.

    Restored system C cannot endure for at least two generations if intermediate system B endured longer than one generation.

Given this interpretation of Kann’s hypotheses, his statements are not contradicted by cases where C failed to endure although A endured for at least one generation, or B endured less than one generation but C failed. This would apply only if Kann were interested in the sufficient conditions of restorations. However, under both interpretations, statement (4) is not supported by Kann’s own data. The first restoration of the Jewish homeland (the second took place in the twentieth century), the first restoration of democracy in ancient Athens and the restoration of Catholicism in Austria did succeed, although the time gap between the original and restored systems exceeded 40 years.

There are no cases contradicting statement (3), but the real reason may be that he simply did not include sequences starting with original systems enduring less than one generation (like the First French republic of 1792–1799 or Napoleon’s empire). This betrays one of the two defects in Kann’s approach. One is conflation between the defining conditions of restorations (expressed by analytical statements) and causal conditions (described by empirical statements). Under the last interpretation, Kann predicts the failure of belated restorations (those which take place later than the time span of one generation). Under the first interpretation, belated restorations are unthinkable, and the very word ‘restoration’ is misapplied in such cases. Both (3) and (4) are affected by this ambiguity, as Kann dismisses as irrelevant not only sequences starting with enduring original systems that are ‘too brief’ but also those with intermediate systems that endure ‘too long’.

He proceeds in this way (Kann 1968: 404–405) dealing with the case of the restoration of Poland in 1918, which did claim continuity of the Polish–Lithuanian Commonwealth or the Kingdom of Poland as one of its two confederated states. In 1795, it becomes one of Europe’s ‘vanished kingdoms’ (Davies 2012). Actually, it was then restored two times—as a vassal state of Napoleon’s empire (in 1807) and as a vassal state of the Russian Empire (in 1815 by the Congress of Vienna). However, after the suppression of the attempted emancipation from Russian suzerainty in 1830–1831, the Kingdom of Poland existed only as an administrative unit within the Russian Empire.

On both counts, more than one generation had passed by 1918. If Kann would have included the restoration of Poland into his set of cases, he would have either grimly predicted the failure of this restoration or balked from his statement that after the passing of one generation, enduring restorations are impossible. Instead, he just dismisses Poland’s case as irrelevant because the time gap between the extinction of the state and its restoration is deemed as ‘too large’, without providing the general reasons for how to distinguish between real (although possibly failed) and only notional restorations.

I accept Kann’s assumption that the capability of the restored system to survive and endure is the most important manifestation of its success. The importance of endurance derives from the paramount aim of restoration to establish a system that will be safeguarded against the recurrence of revolution. I also accept his other assumption that demographic processes in general and the change of generations in particular are key parts in solving the puzzles of restoration. However, I draw a strict distinction between the definition of restoration and the assessment of its success and draw a distinction between the endurance success and performance success of restorations.

Secondly, Kann’s set of cases is too heterogeneous in the temporal and substantive sense. Due to this heterogeneity, it would be ill-advised to simply drop the deterministic interpretation of his theory and expand the population of its cases to arrive at statistically significant results. Kann’s set is too heterogeneous in the substantive sense because his cases include restorations of empires, political regimes, dynasties and even one case of a cultural restoration (the counterreformation in Austria). Other examples of such restorations may be the failed restoration of paganism as the state religion by the Roman emperor Julian (361–363), remembered for this reason in the Christian tradition as Julian the Apostate, as well as the restoration of liberalism (in the guise of neoliberalism) as the hegemonic Western ideology since the 1980s.

In temporal sense, it is too heterogeneous by including cases from four different epochs in Western history: ancient, medieval, early modern and modern. Kann does not take into consideration the fundamental change in Western sociopolitical vocabulary and thinking, brought about by the French Revolution. This change was influentially analysed by the German historian Reinhart Koselleck in his ‘threshold time’ (German Sattelzeit; literal translation ‘saddle time’) theory between 1750 and 1850, during which the language of modernity emerged out of the language of ‘ancient Europe’ (Koselleck 2004 (1979)).

The modern meaning of restoration as a distinctive phenomenon, as opposed to revolution, arose only after the drama of the French Revolution. Kann both starts and closes too early, beginning with biblical times and ending with the establishment of the Second German Empire. On account of its too early start, pre-modern restorations dominate his set of cases, which are not distinguishable from pre-modern revolutions. Likewise, because of its too early closure, Kann leaves out the entire twentieth century, marked by numerous cases of democracy restoration during the second and third waves of democratisation (Huntington 1991).

Most importantly, Kann did not live long enough to witness the post-communist restorations, which refute his pioneering generalisations. Apparently, successful restoration of the Baltic states after some 50 years is the most conspicuous refutation of his theory (pace the prediction, implied by his theory, that their independence will not endure until 2070) because the original systems in the Baltic states endured for only 22 years (1918–1940), while the intermediate systems period lasted for 50 years. Because of the shift in the case population, Kann’s conceptual framework needs to undergo a fundamental overhaul. In its original form, it cannot accommodate post-communist restorations because it commits to define them out of existence because of the too large (40–70 years) time gap separating the original and restored systems or to predict their failure at some time between 1990 and 2070, because they eventuated too late to succeed.

As a first step in this overhauling, I draw a distinction between the endurance and performance success of restorations. The endurance success of a restored social system means that it endures longer than its predecessors (intermediate and original regimes). Performance success means that it increases human wellbeing more than its predecessors and is thus progressive in the absolute (increasing rather than decreasing wellbeing) and relative (accelerating the increase of wellbeing) sense. The time for making the ultimate judgement on the endurance success of post-communist restorations will arrive when restored post-communist states and economic or political regimes will outlast their predecessors (by 2040–2060). However, their performance success (or progressiveness) can already be assessed for each period since the start of restoration.

Assessing endurance success, instead of accepting Kann’s proposal of an absolute measure (70–80 years as a uniform waiting time for all restorations), I prefer to use a relative measure, where the minimal duration required for a restored system to qualify as an ultimate success is relative with respect to the duration of the original and the intermediate systems. This idea is expressed by the:

  • CRES (criterion of restoration endurance success): Restored social system (C) is completely endurance successful if it endured longer than original social system A and intermediate social system B. If the restored system endured longer only than original system A or intermediate system B, it is only partially endurance successful.

CRES implies that social systems where the last revolutions produced relatively enduring intermediate systems, as in the case of communist dictatorships that lasted for more than one generation, take a longer time to prove their lesser susceptibility to a new revolution. Generally, it is also a greater challenge to achieve endurance success when restoring long-lasting social systems than those that lasted only briefly (but long enough to educate their loyalists). If original system A lasted for a shorter time than its post-revolutionary successor B, then by the time the restored continuator system outlasts the original system, partial success of restoration can be celebrated. For the restored Baltic states this was 2012, when they had already outlasted the interwar Baltic states, which endured for only 22 years. However, even before that they could celebrate partial success of the restoration of their democratic regimes, which in interwar Lithuania endured for only 6 years (1920–1926), 14 in Latvia (1920–1934) and for 15 years in Estonia (1919–1934).Footnote 1

Lastly, I draw a distinction between two varieties of restoration: type and token. This step is suggested by an exploration of how restorations are conceptualised in two fields of expertise (protection of cultural heritage and protection of natural heritage), where restoration is an actual profession. Next chapter presents what exponents of the theory of social restorations under construction can learn from these established fields.