A routine visit to grounds in our care, early September. I had read the morning's account before driving over, and it was an unremarkable night. Two areas had been watered in the small hours. The rest had been left alone because the soil was already holding what it needed, and by late morning rain had arrived and the system had stopped everything anyway.
Nothing to see, in other words. Then I took some photographs, and there was something to see.
The middle was right
The turf across the main sweep of those lawns was exactly what the probes had been describing all week. Green, upright, even in colour, no wilt, no footprinting, no blue cast. Ground that was neither thirsty nor sodden. Every decision the system had made overnight, including the several it made to do nothing at all, was borne out by the grass.
That is worth saying plainly, because it is the ordinary case and the ordinary case rarely gets written down. The instrument agreed with the eye.
The edges were not
Along both edges where the lawn meets the paving, and again where it runs up against a retaining wall, there is a band of grass roughly a hand's width across that is thinner and paler than everything behind it. It is on every hard boundary and only on the hard boundaries. It is not disease, which arrives in patches with defined margins. It is not the shape of a failing sprinkler, which would show as an arc or a wedge rather than a stripe.
It is simply the edge. Stone and concrete take the day's heat and give it back into the soil beside them well after the sun has gone, and the ground in that band dries faster than the ground two paces in. On top of that, sprinkler patterns almost always fade before the boundary they are aimed at, so the driest strip of a lawn is often the least watered strip as well.
Why the probes said nothing
A probe measures the ground it is buried in. That is the whole of its virtue and the whole of its limitation. Ours sit in the body of a lawn, because the body of the lawn is what we are there to keep alive, and because a probe in an odd corner would govern the water for ground it does not represent. We learned that lesson the hard way at another estate, and changed how the system weighs its probes because of it.
So nothing we have installed is standing in that band by the paving, and nothing we have installed had an opinion about it. The system was not wrong, and it was not blind in any way it could have avoided. It answered the question it was asked, accurately, about the ground it can feel.
The tempting wrong answer
The obvious response is to water those zones for longer. It is also the wrong one, for the same reason it was wrong when we found a dry corner on another lawn earlier in the summer. Running the whole zone harder would soak the nine tenths that are already correct in order to chase a tenth the water does not physically reach. No amount of extra minutes fixes a problem that is really about where the water lands.
The honest answers are duller. A head adjusted or added so the pattern actually carries to the edge. Or, in some gardens, a border planting that belongs against hot stone rather than turf that never will. Or, sometimes, accepting a slightly thinner edge as the price of a paved terrace and choosing not to spend water on it. Which of those is right is a conversation with the owner, not a decision for a controller.
What the measuring was actually for
Here is the part I keep coming back to. The reason that band took a minute to diagnose rather than a season is that everything else had already been ruled out, in writing, before I arrived.
The lawn was not underwatered, because the root zone had been at target for days and the record says so. It was not stressed by a missed run, because the account lists what ran and what did not and why. It was not a failing zone, because the ground either side of the band is in good order. Strip all of that away and there is only one explanation left standing, and it is the edge itself.
Without the record, that same band is a puzzle. Is the lawn dry. Is a valve weak. Is something wrong with the soil there. You would water it and wait and look again next month, and the month after, and quite possibly never arrive at an answer, because a lawn watered on a timer produces no evidence of its own condition.
Why we still walk the ground
None of this argues for fewer visits. It argues for the opposite.
Instrumentation is very good at the questions it is pointed at and silent about everything else. It knows the root zone and knows nothing about the edge, the mower height, the thinning under a hedge, or the fact that a downpipe has started discharging across a corner. An hour on site sees all of that in a glance and none of it will ever appear in a database.
What the system changes is not whether that hour is needed. It is what the hour is worth. Arriving already knowing that the watering is sound, with the whole-lawn explanations eliminated, means the time goes on the things only a person standing there can notice. Turning up blind, the same hour would have been spent guessing at exactly the questions the probes had already answered.
That, I think, is the right ambition for an autonomous system on private grounds. Not to remove the gardener from the garden. To make sure that when the gardener is in it, none of that attention is wasted on a question the ground has already answered.