Six weeks, then three, then six hours.
The harvest countdown is the only clock on this site that gets less precise the closer it gets, and then suddenly becomes exact. That is not a bug in the interface. It is what harvest is.
Harvest, on patch B2.
Winter wheat on Curdle Hill, drilled on the fourteenth of October. The date on that clock has moved eleven days since May and it will move again.
How the window narrows.
A ten-week window. "Next August" is the honest answer and anything more precise is invented.
Eight weeks. Still governed almost entirely by what spring does.
Six weeks. The crop’s own timetable is now visible and the forecast starts to matter.
Four weeks, and the first honest estimate of what will actually arrive.
Three weeks, and daily moisture readings begin.
The forecast and the moisture agree, or they do not. Nobody is told anything yet.
Moisture is read. Under about eighteen per cent and the decision is made.
The notification goes. The combine is in the field by seven.
Why six hours is the best anybody can do.
The decision to cut is a moisture reading against a five-day forecast, taken at dawn. Above about eighteen per cent and the grain has to be dried, which costs money and can cost more than the difference between cutting today and cutting on Thursday. Below it and the machine goes.
That reading cannot be taken the night before, because dew happens. It cannot be taken three days out, because the forecast is not that good. Every farm in this country works to the same six hours and pretending otherwise on a website would simply mean sending a confident notification that turned out to be wrong.
So the promise is six hours, the notification carries the moisture figure that triggered it, and in two seasons of doing this we have sent one that was wrong — the machine broke down at eight in the morning and the field was not cut for another four days.
What the countdown says at each stage.
| When | What the screen says | Confidence |
|---|---|---|
| October to February | Harvest, next August | A season, not a date |
| March to April | Early to late August | Eight weeks |
| May | The second and third weeks of August | Six weeks |
| June | Around the fourteenth of August | Four weeks, and moving |
| July | A three-week window with a most-likely day | Three weeks |
| The last four days | A date, with a caveat about the forecast | Two days |
| 06:00 on the day | Today. The combine moves at seven | Exact |
Six hours
The final notice, and the best anybody in this country can offer.
Eighteen per cent
The moisture number that makes the decision.
Ignores quiet hours
The only thing on the estate that does, and it has its own switch.
And it is on camera
The big field camera runs through the whole of it.
The countdown, questioned.
Because we would be guessing, and a guess presented as a date is worse than a window presented as a window. The window is real information; the date would not be.
The produce waits. Nothing perishes at the point of harvest — grain goes into store and is milled to order — so a member on holiday loses nothing except the six hours of watching.
Once. The combine broke at eight in the morning and the field was not cut for four more days. We sent a second message saying so, which is the only thing to do.
Yes, on the big field camera, and about four hundred people did last August at seven in the morning.
Watch the field it happens in.
The big field camera, running whether anybody is looking or not.
