I dispute Steve's suggestion that 100 degrees for boiling point is arbitrary without any practical use.
I dispute your disputation of Steve's suggestion. Why 100 degrees between freezing and boiling? Why not 10 degrees? Or not 1 degree? Why freezing? Why not the temperature difference between maximum density and boiling? It's arbitrary.
Freezing and boiling are things we observe water doing on an almost daily basis. These other things are not. It could just as well be 1 degree, or 10, or 1000, but a lot of people are unhappy using either large numbers or decimal fractions (why the hell do we use percentages, for example?). The degree was presumably chosen by Celsius, Fahrenheit, etc., to be a unit that represents a, well, degree
of change which is neither too small to be normally discernible nor so large that one routinely needs to specify fractions of a degree. With the limited accuracy of instruments at the time, specifying changes to a precision smaller than a degree might have given the false impression that they were accurate to that many figures.
To answer Godslayer's point. Yes, the number of degrees between freezing and boiling points could be anything at all. I could memorise freezing and boiling point on the Rankine scale if I liked. Just like I had to memorise 16 ounces to a pound, 14 pounds to a stone, 8 stones to a hundredweight, 20 hundredweights to a ton, 51
yards to a rod, pole or perch and so on. But when I'm calculating something I've got more important things to worry about than conversion factors between units and scales. That's why the metric system, and later the SI system of units, were devised to make all calculations where possible simple powers of 10 without arbitrary scaling factors when combining units (e.g. 1 joule = 1 newton applied over 1 metre).
To anyone who thinks this isn't an issue, I refer them to the fate of a certain Mars probe ...