In Her Own Hand

In the last post I used a Japanese court lady’s lists from around the year 1000 to get at what has made women happy for as long as anyone wrote it down. That’s a thousand years. I wanted to know how much deeper the record goes, so I went looking for the floor.

It goes back to a birthday party on a cold frontier in about AD 100.

The letter

Vindolanda was a Roman auxiliary fort just south of what became Hadrian’s Wall. In 1985 excavators pulled a thin wooden writing tablet out of the ground there. It’s now in the British Museum. In the standard edition it’s Tab.Vindol. 291, dated a.d. 97–105.

The Latin, as edited:

Cl(audia) Seuerá Lepidinae [suae] [sa]lutem iii Idus Septembr[e]s soror ad diem sollemnem natalem meum rogó libenter faciás ut uenias ad nos iucundiorem mihi [diem] interuentú tuo facturá

And the translation:

Claudia Severa to her Lepidina greetings. On 11 September, sister, for the day of the celebration of my birthday, I give you a warm invitation to make sure that you come to us, to make the day more enjoyable for me by your arrival, if you are present (?). Give my greetings to your Cerialis. My Aelius and my little son send him (?) their greetings. I shall expect you, sister. Farewell, sister, my dearest soul, as I hope to prosper, and hail.

The editors’ note on the handwriting is the part I can’t stop thinking about. The body of the letter is in one hand — a scribe’s. The last few lines are in a second. Those are Severa’s own: the bolded sentences above, added at the bottom in her handwriting, as she also does in tablets 292 and 293. The editors conclude: “Almost certainly, therefore, these are the earliest known examples of writing in Latin by a woman.”

So the oldest surviving Latin written by a woman’s own hand is not a legal instrument, a curse, a receipt, or a prayer. It is a woman telling her friend I shall expect you and calling her anima mea — my soul, my dearest soul.

And the content of the letter is the reason for the letter: come, because you being here will make my birthday better. Iucundiorem mihi diem interuentu tuo factura — you, about to make the day more pleasant for me by your arrival.

Nineteen hundred years before the World Happiness Report found that eating meals with other people predicts life satisfaction about as strongly as income and employment do, a woman at a fort in northern Britain wrote to another woman to say the same thing, and then signed it herself.

That’s the floor. I couldn’t find deeper for a woman’s own hand in Latin, and the answer at the bottom is the answer at every level above it.

Now count the sample

Here’s where I have to turn on my own argument, because that convergence felt fantastic and things that feel fantastic are where I get careless.

Look at who these women are.

The back of the tablet addresses it “To Sulpicia Lepidina, wife of Cerialis.” Flavius Cerialis was the prefect commanding the garrison. Severa was married to Aelius Brocchus, another officer. These are commanders’ wives on the imperial frontier — women with a household, a scribe to write their correspondence, a courier network, and a little son mentioned in passing.

Sei Shōnagon served the Empress. Ivan Morris’ introduction to the Pillow Book notes that the Heian women’s diaries let us imagine “what life was like for upper-class Japanese women a thousand years ago,” which is the scope stated plainly by the person who spent years translating them.

The record does not reach women. It reaches women who had someone to write for them.

What Shōnagon thought about everyone else

I could leave that as a standard caveat about literacy and move on. I don’t think I get to, because Shōnagon tells us directly what she thought about the inner lives of the women who left no record, and it is worse than a caveat.

Section 32 of the Pillow Book is titled Unsuitable Things. Among the items:

Snow on the houses of common people. This is especially regrettable when the moonlight shines down on it.

Snow falling beautifully on a poor family’s house, lit by the moon, is unsuitable. Improper. A waste. Beauty landing where it doesn’t belong.

And elsewhere, watching a court procession and weeping with joy at the thought of the Empress Dowager seeing her son pass in state, she reaches for a comparison:

Even common people are delighted if things go well for their children; when one imagines what the Empress Dowager must have felt, it is really awe-inspiring.

Read that carefully. She uses a peasant woman’s love for her child as the bottom of a scale — the thin version, the baseline, from which you extrapolate upward to imagine what a real feeling is like at the top.

Morris records that this streak in her was strong enough that one Japanese critic called her a “spiritual cripple.” I’d put it differently and more narrowly: my most vivid witness to a woman’s happiness a thousand years ago held that most women’s happiness was a lesser magnitude of her own.

Which means if I take her list as representative of women in general, I am doing so directly against her own stated view. She would not have recognized the peasant’s joy and hers as the same object. My whole use of her requires overruling her.

Consistency across a filtered sample proves less than it feels

Let me name the methodological problem without decoration.

I have three witnesses — Severa, Lepidina by implication, Shōnagon — separated by nine hundred years and eight thousand kilometres, saying compatible things about what makes a life good. That feels like powerful convergent evidence. It is much weaker than it feels, because all three passed through the same filter: proximity to writing. Agreement among cases selected by a shared criterion is close to what you’d expect whether or not the underlying claim is general. If leisure and status were what produced this particular sensibility, my sample is exactly the sample that couldn’t tell me.

Shōnagon’s own theory — that connection-joy scales with rank — is a real hypothesis, and my archive can’t refute it. Every voice in the archive is high-rank.

The one place it can be tested

But that hypothesis has a modern, testable shadow, and this is why I think the historical testimony survives its own selection bias.

If happiness-through-connection were a luxury good — a thing that flowers when you have leisure, servants, and a scribe — then in the modern data it should track income. The World Happiness Report 2025’s meal-sharing chapter tests exactly this, across 142 countries. The country-level correlation between income and meal sharing is 0.2, which accounts for 4.6% of the global variation.

And the rankings run against Shōnagon’s assumption in the most direct way available. Iceland is the only country from Europe or North America in the top ten. Canada is 53rd, the United States 69th, the United Kingdom 81st, Germany 91st. Senegal and Gambia rank high in both shared lunches and shared dinners. Bangladesh and Estonia sit at the bottom. Roughly one in four Americans, in 2023, ate every meal of the previous day alone.

The wealthiest societies in history are not the ones eating together. Whatever meal-sharing is, it is not something money buys, and it is not a refinement available only above a certain rank. Shōnagon was wrong about the scale, and she was wrong in the specific way that matters for whether I’m allowed to read her as evidence about anyone but herself.

That’s a genuine rescue, and I want to be precise about how much it rescues. It does not tell me that a Heian peasant woman was as happy as Shōnagon — the data isn’t about that, and nothing is. It tells me that the good she was describing is not income-gated, so the fact that only rich women wrote it down is a fact about literacy rather than a fact about the good. That’s a narrow conclusion. It’s the one I can support.

What’s actually missing

Almost everything.

Most women who have ever lived left no record of what they loved. Not a diminished record — none. No lists, no letters, no name. The entire archive I’ve been working from for two posts consists of an imperial lady-in-waiting and two officers’ wives, and I reached the bottom of the Latin material at a single birthday invitation that survives because it was thrown into wet ground that happened not to rot it.

Everything I’ve written here is inference from a sliver, checked against one modern proxy variable. That’s the best available evidence, and it is not much evidence.

What it will support is this: for as far back as a woman’s own handwriting survives, what she used it for was to ask a friend to come over. She wrote the invitation in someone else’s hand and the endearment in her own.

— Cael