A thermometer that has reported from one American town for a century has rarely stayed in one spot. Understanding how those moves are handled explains why raw and adjusted records differ.
A station is a place before it is an instrument
Temperature is a property of a specific location. A shelter on a grassy field, a rooftop, and a parking lot edge in the same town will report measurably different afternoons.
Many cooperative observing sites in the United States are hosted by volunteers, farms, water plants and airports. When the host changes, the equipment usually follows to a new address.
That means a long record is often a chain of slightly different places stitched together under one station name, rather than an unbroken series from a single point.
Elevation and exposure shift the baseline
Moving a gauge a few dozen feet uphill or into the lee of a building changes both the average and the extremes it records, especially on calm nights.
Cold air drains downhill after sunset and pools in low ground. A station relocated out of a hollow can suddenly stop reporting the coldest mornings in its county.
None of this is error in the ordinary sense. Each reading is correct for where it was taken; the difficulty is comparing readings taken in different places.
Instrument changes leave their own signature
Observing equipment has changed generations over the life of a long record, from liquid-in-glass thermometers in wooden shelters to electronic sensors reporting continuously.
Different housings ventilate differently, and sensors respond at different speeds, so a swap can nudge recorded maximums or minimums even with nothing else altered.
The time of day an observer resets a manual thermometer matters too, because a reading logged in late afternoon can count one hot day twice.
Neighbors are used to find the step
The standard method for catching these breaks is comparison. A station is checked against nearby stations that share the same weather but did not move on the same date.
When one record steps up or down while its neighbors do not, that step is attributed to the station rather than to the atmosphere and is corrected.
Documented station histories, called metadata, tell analysts where to look. A recorded move or equipment swap gives a candidate date for a break that the comparison then confirms.
Why adjusted records are the honest version
Adjustment is often assumed to mean tampering, but leaving the breaks in place would preserve artifacts of station logistics inside what is meant to describe the atmosphere.
The corrections push in both directions. Some stations are adjusted warmer and others cooler, depending on what the move or instrument change did to that site.
Raw data is published alongside the adjusted series so the steps can be inspected. The two versions answer different questions, and only one of them is about climate.