Picture a man fourteen months into a sentence. The induction is long over. He knows which landing is loud at night, which officer will answer, which letters stopped coming. In the Ministry of Justice's filing system, he has quietly crossed a line. He is no longer new. He belongs to the band marked "Over one year".

That band is where the numbers go quiet. For 2024, the most recent year in the published table, the figure recorded against "Over one year" is 22.0. (Source: Ministry of Justice, Safety in Custody -- Deaths_in_prison_custody_1978_to_2024_accessible -- Table_1_3) That is the whole entry. One line, one value, for everyone past their first twelve months inside.

Think about what that single cell is being asked to hold. A man thirteen months into a short sentence and a man thirty years into a life term sit in the same box. Different wings, different ages, different health, different odds. They are counted together because, past the twelve month mark, the statistics stop distinguishing between them.

The most likely reading of why the department splits these figures by time served at all is that time served is believed to change the risk. If length of custody did not matter, you would not build a table around it. The early period gets the fine-grained treatment. Everything after year one gets a single figure of 22.0 for 2024.

Here is my opinion, and it is only that: for the group that makes up much of the prison population, one undifferentiated line is not enough. People serving long sentences age inside. They develop the illnesses that come with age, in buildings that were not designed for frail bodies. A number that lumps year two together with year twenty cannot tell you whether that is going well or badly.

Now the honest caveat, because it is a big one. From this extract alone you cannot be certain what 22.0 counts. The decimal point is a clue that it may be a rate or an average rather than a plain headcount of people, and those are very different things: a rate falling while a raw count rises is entirely possible, and vice versa. A single year also tells you nothing about direction. Without the surrounding years, 22.0 could be the best figure in a decade or the worst, and this story cannot tell you which. Worth remembering too that 2024 is not now. These are the most recent published figures, not this month's.

What the number does establish is that the Ministry of Justice knows how long people had been inside when they died, and publishes it. That is more transparency than many parts of the state manage. The prison estate is one of the few places where the government is fully responsible for whether someone survives the year, and it does put a figure on the outcome.

So take the number for what it is. Not a headline. Not a trend. A line in a spreadsheet covering people who were a year or more into their time, and who did not come out. If you only read one cell of a government table this week, that one is worth a moment.