Original by Don Koks, 2022.
What is the correct way to write the names of SI units? This is not a frequently asked question in physics; but it should be, because many physicists and many more non-physicists get it wrong.
The SI system has two clear and simple rules for how to write its units. The first rule is that all units, when written as a word, are treated as common nouns. In English, common nouns (such as "dog" or "table") are written with an initial lower-case letter—as opposed to proper nouns (such as names of people), which take an initial upper-case letter. So the correct way to write the SI unit of force in English is "newton", not "Newton".
The second SI rule pertains to abbreviations of units. When a unit is not someone's name, its abbreviation is written in lower case. For example, a metre is not someone's name, and so its abbreviation is "m". When a unit is someone's name, the first letter of its abbreviation is written in upper case. A newton is named after Isaac Newton, and so the first letter of its abbreviation (which happens to be just a single letter) is capitalised; hence a newton is abbreviated to "N". Similarly, a watt (named after James Watt) becomes "W", a kelvin (after Lord Kelvin) becomes "K", and a hertz (after Heinrich Hertz) becomes "Hz".
Abbreviations of prefixes are treated separately, because they are not specifically an SI thing: some are written in lower case ("kilo" becomes "k"), and some in upper case ("mega" becomes "M"). Note that these prefixes have unique, well-defined meanings; for example, "kilo" means 1000, and not 1024. In the early years of computer science, computer scientists routinely redefined kilo to mean 1024, because that suited their binary world of powers of 2. They eventually came around to the fact that this is not the way to do science: one cannot redefine such a well-established quantity as "kilo" to suit one's own taste. Hence nowadays—since 1999 in fact—when computer scientists want to denote a factor of 1024, they designate that (in theory at least!) as "kibi", made from the words "kilo" and "binary", and meaning "the number closest to a kilo that is a power of 2". The same idea generates other prefixes now used in computer science, such as mebi (= kibi2) and gibi (= kibi3).
What about plurals? Because SI units are common nouns, they are treated as such when pluralised. We write (and say) "2 metres" and "2 newtons". The hertz is an exception: saying "2 hertzes" just sounds odd, and so we say "2 hertz". Note that we also say (and write, longhand) "2 kelvins", not "2 degrees Kelvin", nor "2 Kelvin".
And what about pronunciation? When you vocalise prefixes, keep them distinct from the unit they precede. For example, millimetre is pronounced "milli meeter", micrometre is pronounced "micro meeter", and kilometre is pronounced "kilo meeter". The latter is not pronounced "klommitter"—after all, no one pronounces millimetre as "mlimmitter". Ending a word's pronunciation with "mitter" is reserved for measuring devices. The pronunciation of these does end in "mitter", such as thermometer, barometer, and micrometer. Compare the latter's stress on the second syllable with the stress on the first syllable in the unit "micrometre".