Chapter 1. Counting time

Measuring a duration and placing a date are two distinct operations, and confusing them is the first source of error. Measuring a duration means answering "how long". Placing a date means answering "from when do we start counting". The first question is settled with units; the second with a starting point chosen by convention.

The official history syllabus of the French state school system, published in 2026, devotes an introductory theme to these two questions, placed at the head of the last year of the course it describes. It asks that pupils know the division of time into second, minute, hour, day, week, month, year, century and millennium, and that they know the instruments for measuring time. That is the right order: you place nothing as long as you do not know in what unit you are counting.

This chapter installs both. It gives the units, the way to write a century, the way to find the century of a year, the notation of dates before and after the starting point, and the useful reminder that the calendar in use today has itself been corrected, on a known date, for a precise reason.

An event is told in hours: a battle lasts a day, a speech lasts an hour. A history, on the other hand, is told in centuries. This is not a stylistic preference, it is a consequence of the sources: a text from the past rarely gives the hour, often the year, and sometimes only the decade. An object taken out of the ground almost never gives better than an interval of several dozen years.

The reading rule that follows from this is useful everywhere. The further back you go, the larger the unit used, and this does not mean that the people of the time lived more slowly: it means that what is left of them is less precise. A prehistoric date given to within a thousand years and a battle date given to the day are not two different levels of care, they are two different natures of source.

The year is the hinge unit. Below it, you are in the event; above it, in the period. It is also the unit that serves as the basis for the two other units of history, the century and the millennium, which are only packets of one hundred and one thousand years.

In the French convention a century is written in Roman numerals. This is explicitly required by the French official syllabus, which counts writing centuries in Roman characters among the expected skills. It has a practical use: on seeing a Roman numeral in a text, you know at once that it is a century and not a year. English-language books usually write centuries in words or in ordinary figures instead; this book keeps the Roman numeral in the table below, because that is the form the convention it describes uses.

Roman numerals are written with seven letters, of which three are enough for every century known: I is one, V is five, X is ten. Letters written from left to right in decreasing order are added, and a smaller letter placed before a larger one is subtracted. So VII is seven, IX is nine, XIV is fourteen and XIX is nineteen.

The trap is not in the writing, it is in the passage from the year to the century. The number of the century is almost never the number of hundreds in the year: 1789 belongs to the eighteenth century, and not to the seventeenth nor to the nineteenth. The reason is mechanical, and the table that follows makes it visible at a glance.

The counting of years used in the greater part of the world has a dated origin. The Institut de mécanique céleste et de calcul des éphémérides writes it thus: the use of the Christian era was introduced in 532 on a proposal by the monk Dionysius Exiguus. The same institute states that this monk had submitted to the pope, in 525, the Latin formula that names this era, and that the usage would spread first in liturgical texts: it would take several centuries, in the institute's words, for it to become common in society. This starting point then imposed itself far beyond Christendom, because trade, administration and science needed a shared count.

Two ways of naming this starting point coexist, and they designate exactly the same thing. The first says "before Christ" and "anno Domini", usually abbreviated to BC and AD. The second says "before the Common Era" and "Common Era", abbreviated to BCE and CE. No date changes between the two notations: 476 BC and 476 BCE are the same year.

This book uses the second notation, BCE and CE, and it uses it everywhere without ever mixing the two. The reason is that this book is translated and read in countries where the first assumes a religion the reader does not necessarily hold. The French edition makes the same choice for the same reason, and the French official syllabus likewise uses the secular form in its landmarks, alongside the traditional one. Both are met everywhere, and you have to be able to read both.

A year is not a whole number of days. The Institut de mécanique céleste et de calcul des éphémérides, at the Observatoire de Paris, gives the value of the tropical year, that is the interval between two passages of the Sun at the spring equinox: 365.242190 days. The Julian calendar, which Europe began to abandon in 1582, country by country, counts 365.25 days per year: its year is therefore slightly LONGER than the tropical year, and it falls behind the Sun. That is the word the institute uses, writing that the reform of 1582 came to correct the lag that the Julian calendar was building up against the Sun.

The gap looks minute, and it is. The institute puts a figure on it: with a mean Julian year of 365.25 days, the dates of the seasons shift by about 3 days every 400 years, that is by a month every 4000 years. Accumulated, it becomes visible. In 1582 the gap had reached ten days, and the spring equinox fell ten days before the date the calendar assigned to it. The reform known as Gregorian corrected the shift in one stroke: the day after Thursday 4 October 1582 (Julian) was Friday 15 October 1582 (Gregorian).

So that the gap would not build up again, the rule for leap years was modified. The institute puts it thus: leap years are the same as those of the Julian calendar (a year whose number is divisible by 4) except three secular years out of four, those whose number is a multiple of 100 without being a multiple of 400. The year 1900 was therefore not a leap year, and the year 2000 was one.

The reform was not applied everywhere at the same time, and the same institute gives the detail. It was adopted as early as 1582 in Italy, in Spain, in Portugal and in the Catholic Netherlands. In France it was applied in December of the same year, Monday 20 December following Sunday 9 December: the ten days suppressed there are therefore not those of October. In Great Britain it took until 1752, where 14 September followed 2 September. This has a direct consequence for anyone reading an old document: one and the same day may carry two different dates depending on the country where the document was written.

Three moves are acquired from here on, and the following chapters assume them. The first is naming the right unit: a duration counted in centuries is not told in years, and the reverse is true too. The second is going from a year to its century, in both directions, checking by the limits. The third is reading a date in both notations without getting the direction wrong.

A fourth acquisition is less technical and more important. The calendar in use today is the result of a dated correction, decided for a measurable reason, applied on different dates depending on the country. It did not fall from the sky, and it is not the only way of counting years. This idea comes back in the next chapter, then in the chapter on the five periods, where it becomes central.