The hatch is on the north side of the house, behind the boxwood, and it takes a flathead screwdriver to persuade the latch. Inside, on a brick pier eleven feet in from the opening, sits a cheap analog hygrometer I bought in 2019 for about fourteen dollars. Beside it, taped to the pier with duct tape that has now gone brittle and been replaced twice, is a second one, digital, that I use to catch the analog drifting. When they disagree by more than four points I write both numbers down and circle the pair.
House was built in 1927, red clay under it, no vapor barrier for the first seventy-odd years and a partial one now. I bought it in 2018. The log is a black-and-white composition book, one page per year, twelve lines and a margin for weather notes. I am on book two.
January readings run low and boring. 46, 44, 49 across the last three years, taken in the second week each time. February is when things start to move, and March is the month I actually care about, because that is when the ground outside warms faster than the brick and the joists stay cold. March 12, 2024: 71 percent, with condensation beaded on the copper supply line near the water heater. March 9, 2025: 68 percent. Same week, same corner, same screwdriver.
Rain matters less than I expected. A two-inch storm in October barely moves the pier reading if the gutters are clear and the grade still slopes away. What moves it is the difference between outside dew point and the temperature of the masonry, which is why late spring is worse than midsummer even though midsummer air is wetter. My June figures sit around 63 to 66. My May figures, more than once, have hit 74.
Gutter work shows up in the log within about six weeks. After I re-pitched the back run in August 2022, the following spring peak dropped from 76 to 69, and it has stayed in the high sixties since. The dehumidifier I ran through 2021 pulled numbers down four or five points while it was on and then everything went back where it had been the moment I unplugged it, which told me more than the four points did.
Memory is the reason for the pencil. I would have sworn, before checking, that the wet season under this house was August. It is not. August is a middling month. I had confused the misery of standing outside with the condition of the ground under the floor, and the log corrected me flatly, without argument, in my own handwriting from four years back.
A neighbor two streets over keeps something similar for a 1908 house with a stone foundation, and his shoulder-season spike is sharper and about a month earlier than mine. We compare pages in the fall. Neither of us has drawn a conclusion from that yet.
Next reading is due the second week of October. I expect somewhere between 58 and 62, and I will write down whatever it says instead.