Why the reading hour matters more than the gauge does

01

One gauge, one time, one notebook

A rainfall total is not a measurement of the sky. It is a measurement of the interval between two readings, and if the interval wanders then so does the figure. Reading at seven one morning and at six the following evening gives you thirty-five hours in one box and thirteen in the next, and a shower that falls at five in the afternoon lands in a different day depending on when you happened to cycle up the lane.

Since August 2020 I have read at eight in the morning and written the total against the previous day, because the overwhelming majority of what falls in the twelve hours before eight has fallen in the night that belongs to yesterday. Calling it yesterday's rain is a convention rather than a truth, and the only thing that matters about a convention is that it never changes. The notebook has a printed line at the top of every page reminding me which day the figure belongs to, because I got it backwards twice in the first winter.

The other half of consistency is the notebook itself. Figures written on a phone at the plot got lost twice when I changed the device, and figures held in my head until I got home were wrong about one time in five. Pencil in a hardback book at the shed door, then a copy into the spreadsheet on Sunday. The pencil is because biro does not write on damp paper.

• 08:00 reading, credited to the previous day, without exception
• Rain to the nearest 0.5 mm, temperature to the nearest 0.5 C
• Missed readings written as M, never filled in afterwards
• Remarks column for hail, snow lying, and standing water

02

Where the gauge actually stands

The first place I put the gauge was on top of a fence post by the shed, because it was convenient and the gauge would not get kicked. A gauge on a post two metres up catches noticeably less than one at ground level in anything windy, and it catches a good deal more than it should whenever water runs off the shed roof edge in a north easterly. I did not know either of those things when I chose the spot.

It now sits in the open, sunk into the turf so the rim is 300 mm above short grass, with the nearest obstruction more than twice its own height away in every direction except the hedge, which is four metres off and about two metres tall. None of that is ideal and I know it, but an allotment is a small rectangle with things growing on it and the alternative sites were all worse.

Splash was the other lesson. Standing on bare soil, a gauge collects droplets thrown up from the ground around it in heavy rain, and in a summer thunderstorm that can add a couple of millimetres to a figure that already matters. Short grass around the base, kept short, took that away. I mow a circle a metre across with shears about once a fortnight from April, which is the least glamorous part of keeping a weather record.

03

The second gauge that disagreed

In 2022 I put an identical gauge four metres north of the first one, at the same height, purely to see what two instruments that ought to agree would actually do. Over the first six months the totals came out at 341.5 mm and 328.0 mm. The gap works out at four percent over the period, which I could live with, but the daily figures were much further apart than that and the direction changed with the wind.

On days with rain falling more or less straight down the two gauges were usually within half a millimetre. On days with strong wind from the south west, which is most of our wet weather here, the northern gauge read consistently low, because it sits a little closer to the hedge and the hedge does something to the airflow that I do not fully understand. The largest single day difference was 4.5 mm against 7.0 mm on a February afternoon of driving rain.

The useful conclusion was not that one gauge is right. It is that a single figure from a single allotment gauge carries an error of roughly half a millimetre on a quiet day and considerably more in wind, and that I should stop quoting daily totals to a precision they do not deserve. Monthly totals are where the record starts to mean something, and the two gauges have never differed by more than eight percent over a month.

04

Reading a maximum and minimum thermometer honestly

The thermometer lives in a small slatted wooden screen I made from offcuts, painted white, with the bulb at 1.25 m and the door facing north. Before that it hung on the shed wall, which on a sunny March morning reads four or five degrees above anything the air is doing. Every figure from that period is in the notebook with a line through it.

The reading itself is simpler than the rain but easier to fumble. I read the minimum first, because that is the number I usually care about, then the maximum, then reset with the magnet, and I check that the reset actually happened before I close the door. Twice in the first year I recorded a minimum that was really the previous day's, because I had forgotten the reset and the index had not moved.

The soil level thermometer, lying on short grass under the screen, is the one that tells me about frost on the plot. It regularly reads three to four degrees below the screen minimum on a clear still night, and on 16 April 2024 it read minus 2.5 C while the screen minimum was plus 1.0 C. If I only had the screen figure I would have recorded that night as frost free, and the potato tops that got blackened would have gone into the remarks column as a mystery.

05

What I stopped writing down

The first version of the notebook had nine columns, including wind direction estimated from the hedge, cloud cover in eighths, and a subjective note on how the soil felt. Two of those three went within a year. Wind direction guessed from the movement of a hawthorn is not data, and my eighths of cloud at eight in the morning tell you mostly about whether I had had coffee.

What stayed are the four numbers that come off instruments, the date, and a remarks column in plain language. That column has turned out to be the most valuable part of the whole record, because it is where I wrote things like ground still frozen at 50 mm and cannot dig, or third day of standing water in the bottom beds. Those sentences anchor the numbers to what the plot was actually like.

The record is now six years long and dull to look at, which is the point. It took about four minutes a day and it has settled at least a dozen arguments I was having with my own memory, usually in favour of the notebook.