The chronicles' hard years, checked against the tree rings

What the count shows

The published compilations give 157 calamity records covering 111 distinct years (a year can appear several times, with different troubles). The years written down as drought (46 records) lean dry in the rings: the middle summer of their run was drier than 62% of the summers of 1450–1800, and if the years were drawn at random from the same period, a mean that dry would turn up about 3% of the time. It is a modest signal: this page runs fifteen such tests (five kinds of trouble × three lags), and a strict correction for multiple testing would erase it. Famines (12) are indistinguishable from chance (66%), as are floods (37, tested on the wet side: 29%) and locusts (16, 66%). The hard winters (46), the negative control — a summer index cannot see them — show nothing in their own year, but the summers before them tend to be dry, a result printed unexplained below, because it may be chance among many tests or an artefact of how annals pin a winter to one year or the next.

The famines the trees refuse

1534 is the year of the great famine in the Făgăraș district, blamed by the sources on drought. In the rings, the summer of 1534 was wet: wetter than 96% of the summers between 1450 and 1800. 1589, a year of “dearth and great famine”, is the same: wetter than 98% of summers. Those are the two wettest summers among the twelve famine years in the table, not typical cases: the middle famine summer sits slightly on the dry side (drier than 56% of summers). Which is why the conclusion is narrow: with twelve years, the test would not see even a moderate drought effect. What can be said is that the count does not find summer drought as the famines' common cause, and that two famous famine years flatly contradict the cause their sources name. A harvest can also be lost to too much water, to war, to requisitions, or to the winter before; the trees judge only one of those causes.

The driest summers, mentioned and unmentioned

The driest summer of the period in the rings, 1686 (-4.8), is also written down as very dry in the Transylvanian sources. But the list also holds extreme summers for which the compilations used here record no drought: some appear instead as hard winters or rainy years, others not at all. The values in this table are the mean over all the Romanian lands; in the large table each event is judged on its own land, so the same year can carry a different value there (1686: −4.80 here, −5.15 for Transylvania). “Unmentioned” here means unmentioned in the four compilations behind the table, not unmentioned anywhere: Andrea Kiss has already compared Carpathian-basin documents with this same tree-ring reconstruction, and the drought of 1516–1517, missing from the sixteenth-century Transylvanian list, is documented in her work.

The calendar trap

The Moldavian chronicles count years from the creation of the world, with the year starting on 1 September: chronicle year 7152 runs from September 1643 to August 1644 in modern counting (subtract 5509 for September–December, 5508 for January–August). A harvest lost in one summer becomes a famine the next spring, sometimes under a different chronicle year. So every match on this page was also tested one year earlier and one year later: for droughts the signal is strongest at the exact year, which supports the compilations' dating for that class. For winters the same test points the other way — the signal sits a year early — so the rule cannot be stretched across all classes. Where the sources themselves disagree about the year — the great drought put by some in 1473 and by others in 1474 — the row stays in the table with the disagreement written beside it and stays out of the count.

How well the trees see, where thermometers can check them

The tree-ring reconstruction was set against measured June–August rainfall at five stations (Bucharest-Filaret from 1895, Cluj from 1923, Iași, Sibiu and Băneasa with usable series from 1961). Correlation over each station's own period runs from 0.31 to 0.63; when one half-period is left out of the fit and used only for checking, values run from 0.06 to 0.73, by station and by half. In plain terms: the rings catch the direction of a summer, not a safe verdict on any single year. 1816, the “year without a summer” after Tambora, shows in the rings as a very wet summer (+2.11, wetter than 97% of the summers of 1500–1900). The summer after the great winter of 1708–09 was wet too (+1.39, wetter than 90% of the summers of 1500–1900): a summer index cannot judge the winter itself, and it carries no dry trace of it either. Europe's famous drought of 1540 is, over the Romanian lands, an almost ordinary summer (-0.63): its centre lay further west.

The unexplained result, in numbers: the summers before hard winters average -0.97 in the rings, and a mean that dry would arise by chance about 0.1% of the time — the smallest percentage among this page's fifteen tests, produced by the control class itself. No conclusion on this page uses it. And a note on what the index measures: it combines rainfall with evaporation, so a cold summer can read as “wet” without heavy rain — one way to read 1816. Throughout this page, “dry” and “wet” mean this index, not rainfall measured directly.

Your year's ring

Type a year between 1450 and 2025. Up to 2012 the rings answer; after 2012, measured station rainfall, labelled as such.

Method and sources

The event years come from four published compilations: Gaceu et al. 2024 for sixteenth-century Transylvania; the climate history of Transylvania in 1645–1715 from German-language sources (Climate, 2022); Andrea Kiss's documentary droughts of the Carpathian basin; and Elena Teodoreanu's synthesis after Cernovodeanu & Binder (1993). Each year was classified by what its source says, before looking at the rings; a second reader reclassified 24 quotations blind and agreed in all 24 cases. The tree side is the Old World Drought Atlas (Cook et al., Science Advances, 2015): a summer drought index reconstructed from tree rings alone, on a half-degree grid, averaged over each event's own land. Agreement was tested against shuffled years (4,000 draws) and at ±1-year lags. The large limits: the index sees only summer drought; of the 157 records, 138 concern Transylvania, 16 the Romanian lands as a whole, and only 3 name Moldavia or Wallachia specifically — so the count says more about Transylvanian sources than about the Moldavian chronicles; the absence of a mention is not proof of the absence of the event.

• Cook, E.R. et al. (2015). Old World megadroughts and pluvials during the Common Era. Science Advances 1(10). Data: NOAA NCEI, OWDA v1.0.
• Gaceu, O. et al. (2024). Reconstruction of climatic events from the 16th century in Transylvania. Frontiers in Climate 6:1507143 (CC BY).
• Climate History of the Principality of Transylvania during the Maunder Minimum Years (1645–1715 CE). Climate 10(5):66, 2022 (CC BY).
• Kiss, A. (2015). Droughts, dry spells and low water levels in medieval Hungary (and Croatia) I. Journal of Environmental Geography 8(1–2).
• Teodoreanu, E. Preliminary observations on the Little Ice Age in Romania. Present Environment and Sustainable Development.
• Cernovodeanu, P. & Binder, P. (1993). Cavalerii Apocalipsului [The Horsemen of the Apocalypse: natural calamities in Romania's past, to 1800]. Silex (cited through the compilations above).
• GHCN-Daily measurements (NOAA NCEI): Bucharest-Filaret, Bucharest-Băneasa, Cluj-Napoca, Iași, Sibiu.