Fresh Sheets

20 articles
Show all %d

History as tinchalice covers it is the study of how people organised power, moved goods and people, and remembered what happened — worked out from the surviving evidence and dated with as much precision as that evidence allows. That means three habits run through everything here: a hard separation between what a source says and what a historian infers from it; attention to periodisation, because the dates you choose change the story you tell; and a refusal to smooth over the places where the record is thin, contested, or simply absent. The rest of this page is about how those habits play out in practice, and what a reader gives up by adopting them.

How a claim about the past becomes usable

A usable historical claim names its evidence and its date. That sounds like a low bar. It eliminates most of what circulates as history.

Take the death toll of the American Civil War, fought between the Union and the Confederacy from 1861 to 1865. The figure most often quoted now is 750,000 dead, a number the U.S. National Park Service has used since 2023. That number is not a count. Nobody counted. It is a demographic estimate built by comparing census cohorts before and after the war and inferring how many men of military age failed to appear in the later count, then adjusting for civilian deaths, migration, and the ordinary undercount of a nineteenth-century census. Earlier generations of historians used a figure closer to 620,000, and the revision upward came from better statistical method, not from new documents. So when you read "750,000," you are reading an argument about method as much as a fact about the war. The honest version of the sentence carries the attribution; the dishonest version drops it and lets the number look like a body count.

This is the single most useful skill in reading history, and it is boring. Every figure you meet has a provenance: a census, a tax roll, a chronicler with a patron, a modern statistical model, or a journalist repeating another journalist. Ask which. A medieval chronicler's casualty figures for a battle are usually symbolic — round numbers, biblical echoes, multiples of twelve — and reading them as counts produces nonsense. A parish register is a count, but only of the people the parish priest bothered to record, which in most places and periods means it undercounts the poor, the young, and anyone who died away from home.

The cost of this discipline is that it makes history slower and less quotable. You cannot say "the plague killed a third of Europe" in casual conversation without either footnoting yourself or accepting that you are repeating a figure whose basis you have not checked. The compensating gain is that you stop being surprised. Once you know that most premodern numbers are estimates dressed as records, a great deal of popular history stops working on you.

Periodisation is an argument, not a fact

Every start and end date is a choice, and the choice carries a thesis. The conventional date for the fall of the western Roman Empire is 476 CE, when Romulus Augustulus was deposed. Nothing much happened in 476. The deposition of a child emperor by a Germanic commander was one event in a long process; the empire's tax base, its army, and its political centre had been fragmenting for generations, and in most of the west daily life continued much as before. The date survives because it is convenient and because it flatters a story about a clean break between antiquity and the Middle Ages. People who study late antiquity for a living mostly do not believe in the clean break. They will tell you the eastern empire kept going, that Roman law and Roman bishops outlasted Roman armies, and that the interesting question is not when Rome fell but what "falling" would even mean for an entity that had already stopped being centred on the city of Rome.

The same problem, sharper, applies to the end of the Second World War. The war is conventionally dated 1939 to 1945, with the Pacific War continuing until Japan's surrender, which came on September 2, 1945. That date is real in a way 476 is not — it is the day a document was signed on a ship — but it is still a boundary drawn around a process. It does not mark the end of fighting everywhere in Asia, the end of occupation, the end of the conditions that caused the war, or the release of prisoners. Treating September 1945 as the moment the world returned to normal is a choice that centres the European and Pacific theatres and ignores what was still happening in China, in Southeast Asia, and across the decolonising world.

What you get from taking periodisation seriously is the ability to read a date as a claim. A textbook that opens a chapter in 476 is telling you it believes in a rupture. A book that opens in 400 and ends in 600 is telling you it believes in a transition. Both can be good books. The point is that you should notice which one you are holding, because the dates are doing argumentative work whether or not the author admits it.

The trade-off is real: periodisation is unavoidable. You cannot write or think about the past without chopping it into manageable pieces, and the conventional pieces — antiquity, the Middle Ages, the early modern period, the modern — are the ones your reader will recognise. Using them is a communication decision, not an endorsement. The trick is to use them while saying out loud that they are conventions.

Where the sources run out, and how to read the gap

Most of human history is undocumented, and the undocumented parts are not evenly distributed. They cluster around people without property, without literacy, and without institutions that kept records: women in most periods, enslaved people, the rural poor, children, and the populations of regions that literate empires regarded as peripheral. The absence is not neutral. It is a product of who had the means to write and who had an interest in preserving what was written.

Historians handle this in a few recognisable ways, and knowing them helps you spot when an author is bluffing. One is to read the archive against itself: a tax record compiled by a landlord can tell you what tenants grew and owed even though it says nothing about what they thought. Another is to use material evidence — archaeology, bones, pollen, building remains, shipwrecks — which does not depend on anyone deciding the poor were worth writing about. A third is to reason from later sources backward, carefully, and to say plainly when the reasoning is inference rather than record.

What goes wrong is the fourth approach: filling the gap with plausible reconstruction and letting it read as fact. This is where popular history about the ancient world and the early medieval period most often fails. A sentence like "the household gathered each morning to…" is usually an inference from a single source, generalised across centuries and regions. It might be right. You have no way to know, and neither does the author, unless they tell you. The tell is the absence of hedging: a historian working close to thin evidence hedges constantly, because that is what the evidence requires.

The cost of insisting on this is that a lot of history becomes unsatisfying. You will read a whole chapter and come away with "we don't know, but probably this, in this region, at this time." That is the correct output. The gain is that you can tell the difference between a historian who has run out of evidence and one who has run out of honesty, and those are very different failures.

Dates, calendars, and the problem of converting them

A date in a premodern source is not the same object as a date on a modern timeline. Converting one to the other is an operation with error bars, and the error bars matter.

Three problems recur. First, calendars differ: the Julian calendar in use across Europe until the Gregorian reform drifted against the solar year, so a date recorded in one system does not fall on the same day as the equivalent date in the other. Second, year numbering differs: regnal years, consular years, and years counted from a religious or founding event all have to be converted, and the conversion depends on assumptions about when the year began — which varied by region and by purpose. Third, some events are dated by season or by a religious feast rather than by a day, which means the best you can do is a range.

This is why serious reference works give ranges and why you should be suspicious of any popular account that gives a precise day for something ancient. The precision is usually imported from a single source that happened to record it, without asking whether that source was in a position to know. A chronicler writing two hundred years after the fact, working from an earlier chronicle, is not evidence for a day. He is evidence for what people two hundred years later believed.

For anything within the last few centuries the problem mostly disappears, which is one reason modern history is easier to write and easier to get wrong in a different way: there is so much documentation that selection becomes the argument. A history of a modern war built entirely from government archives will tell a different story from one built from letters, and both will be accurate.

Scale: the difference between an event and a process

Most of the interesting questions in history are about processes, and most of the writing people actually read is about events. Getting the two straight is the difference between understanding a period and memorising its highlights.

An event has a date and usually a decision: a battle, a deposition, a surrender, a law passed. A process has a direction and a duration and no single moment you can point to: the accumulation of a tax base, the slow shift of a language, the build-up of a demographic crisis, the entrenchment of a legal distinction between free and unfree people. The Civil War's 750,000 dead is the outcome of a process — mobilisation, disease, logistics, the decision to keep fighting — that a list of battles will not explain. The fall of the western Roman Empire in 476 is a date attached to a process that had been running for a very long time and did not stop when the date arrived.

Historians who work at the scale of decades and centuries are usually trying to explain processes, and their books are often unsatisfying to read because processes do not have protagonists. Historians who work at the scale of a year or a life are usually trying to explain events and decisions, and their books are more readable but can mislead by making the process look like a series of choices by named individuals. Both scales are legitimate. The mistake is to read a book at one scale as if it were answering a question at the other.

What this costs you as a reader is patience. A process-level explanation rarely gives you a villain or a turning point, and it is much harder to remember. The gain is that you stop being surprised when a country does something its constitution and its economy have been pointing toward for a century. Events are what get reported. Processes are what determine which events were possible.

What the evidence lets you conclude, and what it does not

The test of any historical claim is whether a competent sceptic could check it against the evidence the author cites. That is a lower bar than truth and a much higher bar than most history writing clears. It means a claim can be well-evidenced and still wrong, and it means a claim can be true and still unsupported by the sources the author had.

In practice, this gives you a rough hierarchy of confidence. Inscriptions, administrative records, and contemporary documents produced by someone with access to the facts sit near the top, with the caveat that they were produced for a purpose and encode that purpose. Narrative histories written close to the events sit below them, because they compress and moralise. Histories written long after the events sit below that, useful mainly for what they show about the later period's beliefs. Material evidence sits alongside all of these and is often the only check on them, but it answers different questions: it can tell you what people ate and where they lived and how tall they were, and it cannot tell you what they thought about it.

What this rules out is the confident synthesis. The books that sell best are usually the ones that resolve every ambiguity into a clean narrative, and the resolution is almost always a choice the author has made rather than a finding the evidence forced. That does not make them worthless — a well-argued synthesis is how most people get their picture of a period — but it does mean you should read them knowing where the seams are. The seams are exactly where the author stops citing and starts asserting.

The cost of holding this standard is that it makes you a difficult reader. You will notice when a book's central claim rests on a single source, when a date has been quietly rounded, when a death toll has been repeated without its method. You will also notice how much good history is written under exactly these constraints, hedging and qualifying and still managing to say something. That is the thing worth learning to recognise: not the confident paragraph, but the paragraph that knows precisely how far its evidence will carry it, and stops there.