It's even worse. Even the data that has not been modified is subject to so many conjectures.
17xx - the process is definitely manual. Perhaps, some student got paid one sou a day to record stuff. Now, one day he is ill, or got drunk, or got lazy, and just written down some numbers, or written them in the wrong cell in the spreadsheet. Then also the reading is never right on the dot. Did he round it up or did he round it down? Then the thermometer has been installed "somewhere". How do you translate that to "the temperature 2m above the ground" that is the standard today? Then 1795 comes, the scale changed from Fahrenheit to metric, the thermometer gets moved to a different building, the student is guillotined in 1793, and all of that changes.
Ok, maybe not quite so bad for Paris, but the raw UK data I've seen was full of problems of this sort. Out of 100 years of data from about 20 gauges only about 7 gauges had completely overlapping 20 years.
no subject
17xx - the process is definitely manual. Perhaps, some student got paid one sou a day to record stuff. Now, one day he is ill, or got drunk, or got lazy, and just written down some numbers, or written them in the wrong cell in the spreadsheet. Then also the reading is never right on the dot. Did he round it up or did he round it down? Then the thermometer has been installed "somewhere". How do you translate that to "the temperature 2m above the ground" that is the standard today? Then 1795 comes, the scale changed from Fahrenheit to metric, the thermometer gets moved to a different building, the student is guillotined in 1793, and all of that changes.
Ok, maybe not quite so bad for Paris, but the raw UK data I've seen was full of problems of this sort. Out of 100 years of data from about 20 gauges only about 7 gauges had completely overlapping 20 years.