Ruler Survival in China, Europe, and the Islamic World

Figures from Chen & Lin (2026)

The following figures are reproduced from Chen and Lin, "A Quantitative History of Regicide in China," Ch. 4 in Quantitative History of China (Springer, 2026), pp. 65–108. Open access, CC BY-NC-ND 4.0. Subsample B = mainstream emperors from 221 BCE to 1911 CE (excludes fragmentation-era petty kings, who are in Subsample C). European data from Blaydes & Chaney (2013) and Eisner (2011).

Fig. 4.13. Ruler duration comparison among Chinese (Subsample B), European, and Islamic rulers. 100-year moving averages. China (blue) shows high volatility with no secular trend. Europe (gray) shows the only genuine secular increase. Islam (black) stays flat or declining. Source: Chen & Lin (2026, p. 98), data for Europe and Islam from Blaydes & Chaney (2013).

Fig. 4.10. Regicide rate comparison between China (Subsample B) and Europe. Regicide rate = number of regicides per 100,000 ruler-years. After 700 AD, the two series converge and decline in parallel toward near-zero by 1800. The decline in elite political violence is a shared Eurasian phenomenon, not uniquely Chinese. Source: Chen & Lin (2026, p. 94), European data from Eisner (2011).

I. Three Sources Compared

II. Dynasty-Level Ruler Tenure: A Comparison

The table below compares Huang's stated dynasty averages (from EAST, Ch. 5, pp. 162–63) with averages computed from Wang's individual ruler data (Harvard Dataverse). Differences arise from sample composition: Huang appears to count total dynasty years ÷ number of emperors, while Wang's data records individual tenures that may exclude interregna or co-regencies.

III. Analytical Notes

A. Does Chinese ruler tenure increase over time?

Huang's central empirical claim is that Chinese emperors ruled longer as time progressed, attributing this to the Keju system's pacification of elites. His pre-Sui vs. post-Sui t-test (16 vs. 25 years, p=0.043) is the statistical backbone of this argument (Table 5.1, p. 168).

The problems with this test are severe. Huang uses dynasty-level averages as his unit of analysis, generating one observation per dynasty. His pre-Sui group has 6 observations (Qin, W. Han, New, E. Han, Han-Sui Interregnum, Sui); his post-Sui group has 7 (Tang, Five Dynasties, N. Song, S. Song, Yuan, Ming, Qing). This is a t-test with n=13. All within-dynasty variation is suppressed. Wang Mang's 16-year New dynasty (a single usurper) carries the same statistical weight as the 267-year Qing. When Chen & Lin (2026, Fig. 4.13) apply a 100-year moving average to individual ruler data (Subsample B: mainstream emperors, excluding fragmentation-era petty kings), the upward trend disappears entirely. The China line is wildly volatile, spiking during stable phases (Song, ~27 years) and crashing during transitions (Mongol conquest, ~5 years). There is no secular trajectory.

The result is also fragile to its composition. In the pre-Sui group, the Han-Sui Interregnum (220–581) compresses 361 years of political fragmentation spanning dozens of competing regimes and roughly 90 rulers (per Wang's data) into a single observation with a low average. This one data point does substantial work pulling down the pre-Sui mean. In the post-Sui group, the Qing does the opposite: Kangxi's 61-year reign and Qianlong's 60-year reign inflate the dynasty average to 36 years, the highest of any dynasty, which pulls the post-Sui mean well above where it would otherwise sit. Remove the Qing or split it differently and the significance likely vanishes. A test whose result depends on one observation at each tail is measuring the idiosyncrasies of those observations, not a structural historical process.

There is a further problem of circularity. Huang's grouping variable (pre/post Sui) is chosen to correspond to the mechanism he proposes (Keju, which began under Sui). The cutpoint is the theory. If you moved the boundary one dynasty in either direction, or split at the Song (where Keju became fully meritocratic), the result would change. With n=13, a single reassignment of one observation across the boundary can flip significance. This is not a test that could survive a pre-registration regime.

B. Is China's ruler stability distinctive?

The cross-civilizational comparison (above charts) shows that China and Europe are remarkably similar in ruler tenure throughout the 1000–1800 window, both hovering in the 15–18 year range. The Islamic world is the real outlier, declining from ~16 to ~11 years. If increasing ruler tenure caused stagnation (Huang's argument), Europe should also have stagnated, since European rulers enjoyed the same or greater tenure gains.

Chen & Lin's Figure 4.10 reinforces this: Chinese and European regicide rates converge after 700 AD and decline in parallel toward near-zero by 1800. Elite political violence declined across Eurasia, not uniquely in China.

C. Where the data does show Chinese distinctiveness

Wang's deposal rate data reveals the genuine Chinese anomaly. Chinese rulers went from being deposed by elites 55% of the time (c. 1000) to just 20% (c. 1800). European deposal rates only declined modestly (46% → 30%). This divergence is real and not compositional, since the Wang sample covers 1000–1800, a period of predominantly unified empires in China.

The question is whether this declining deposal rate reflects a deliberate ruler strategy (Wang's "sovereign's dilemma": rulers weakened the state to ensure personal survival) or a byproduct of institutional change (the Keju system pacified elites and fragmented gentry networks, making coups structurally harder regardless of ruler intent). The fiscal data supports the second interpretation: China's tax/GDP ratio fell from ~17% under the Song to ~1% by the late Qing (Wang, Fig. 2.9; Guan, Ma, & Zhai 2026), but the fiscal decline tracks the localization of elite networks after the Tang-Song transition more than any deliberate policy choice by rulers.

IV. Sources