There are three different Latin
roots
test-. None of them has a plausible
etymology, and there is no known connexion between the three. The three roots mean WITNESS, POT, and GONAD. There are no
known origins, but it is natural to speculate on how they might be related.
I'm going to discuss how these words made their way into English derivatives, then I'm going to speculate, and dissect the juicy traditional story about Romans swearing on their testicles.
WITNESS
First to amplify on the meanings. In Latin there was a word
testis, meaning
witness. This could be masculine or feminine. It didn't specifically mean one sworn in court, because it could just mean an
eye-witness.
Related to this were a number of words with obvious English counterparts: testificari 'to give evidence, vouch for, testify; to make public; to call to witness', testamentum 'will, testament', testimonium 'evidence, testimony, proof'.
Less obviously related are the compound verbs: to attest is to witness to, to contest is to call witnesses in a case with someone, and to protest (my little Latin dictionary doesn't contain this one: and drdave confirms it isn't Classical) is to stand forward as a witness.
POT
There were three Latin words
testa,
testu, and
testum, all meaning something like
pot,
potsherd,
tile,
brick,
lid,
shell. I'm using POT as the central meaning of these, but as we don't know the etymology we might be mistaking a developed form for an original form. That is, I don't necessarily mean it began as round 'pot' then spread out to mean 'tile' and 'brick' as well. It might make a difference if the meanings spread in a different order.
My little dictionary says for testa: 'brick, tile; (earthenware) pot, jug, sherd; (fish) shell, shell-fish'. For neuter testu and testum it gives 'earthenware lid, pot'.
The principal development of this root in English is test, which originally meant a crucible, a small pot. The test was what you put a substance in to assay it. From that, test came to mean the process of assaying or the result of an assay, and from that it was generalized to any other examination: the modern sense of 'test'.
In Vulgar Latin testa came to mean 'head': it replaced the Classical Latin caput. That continued to mean head in the abstract sense of chief or leader (capital and chief and chef are related), but in the anatomical sense, the head became the testa 'pot', possibly via an intermediate sense of 'skull'. This is modern French tête, borrowed in English in tête-à-tête.
GONAD
The final Latin word is
testis, masculine only, meaning the male
gonad, and this is definitely related to its
diminutive testiculus. These are borrowed into English as
testis and
testicle.
etymology of WITNESS
The on-line American Heritage Dictionary is quite good on etymology, referring things back to
Proto-Indo-European roots and drawing out unusual connexions, but I think they're sticking their necks out here. Their suggestion (and presumably they're repeating that of some respected Indo-Europeanist linguists) is that
testis WITNESS comes from an older form such as *
trestis, which unpacks into the two roots
tre- '
three' and
st(a)- '
stand'. That is, a
witness is a third person standing by.
Well, that works perfectly as an explanation of the meaning, but the phonetic form is dubious. Why should Old Latin *trestis have lost its R? I don't know of any other examples of this with the root tre-, nor of any general rules for consonant loss in this position. What Old Latin had was a strong initial stress, which did cause quite a bit of phonetic reduction in later syllables and clusters, but not at the beginning.
Another factor in Latin is dissimilation: if there had been a word *trestris or *trerstis it would be easy to account for losing one of the two R's; but that's not the case here.
Also, we'd expect to see similar compounds like bistis, instis, constis, exstis, abstis, substis, superstis, but what we actually have in Latin are present participle forms instans, constans, substans etc. (with stems in -stant-). So *tre-st-i-s looks unlikely in morphology too.
So I find this tre- implausible. But there is no other obvious etymology for testis WITNESS: no other known verbs or adjectives that fit with that sense.
If you've got knowledge of this area (hi, Gone Jackal and Muke) and can point me to reasons why this loss of R is phonetically plausible, I'm interested in hearing. Thanks then to drdave for more discussion after this was posted: I still don't buy it, but here's a bit more about the tre-st- possibility, which I'm going to relegate to an Appendix below because it's getting hairy.
etymology of POT
The etymologist
Skeat suggested that the root
test- POT came from earlier *
terst-. Now in this position, in a cluster after the stress, loss of R is quite natural in
Old Latin. That would mean 'dried, baked', from the root
*ters- '
dry,
bake', so the original sense would be not POT but any
earthenware.
This root is also related to English 'thirst', to Latin torridus 'torrid', and foremost to Latin terra 'earth', earlier *tersa. As well as giving derivatives such as terrestrial, inter (bury), terrace, and terrier (a burrowing dog), this corresponds to Irish tír in their mythological Tír na nÓg, Land of Youth.
However, there is a Persian word tast meaning 'cup'. If this is related to testa, testum then it is the sense of cup or pot that is original, and they are less likely to be related to 'dry, bake'.
etymology of GONAD
The fact that
testis meant both WITNESS and GONAD naturally leads some to suspect that GONAD is a
metaphorical extension of WITNESS. The Chambers and Webster dictionaries just mention 'witness (of
virility)' in passing. But why would you call testicles witnesses? We don't call breasts signs of femininity: we don't
call them such things. We have simpler words.
Euphemistically, the
penis is called
manhood, and
private parts may be called by the form (globes, balls, garden) or the function (her sex), but such an indirect euphemism as 'witness to virility' strikes me as highly implausible.
So the other line of connexion that's been made is the suggestion that witnesses in Roman courts swore clutching their testicles. I am not a historian: I have no idea whether this is true or how much evidence there is for it. Any takers? I have never heard of this testicle testimony story except for the purpose of connecting these two words.
What I do know is that in Genesis 24:9 someone swears by placing his hand "under the thigh" of Abraham, and my old Hebrew teacher told me this was a euphemism for clutching the genitals, and I trust him to be right. Again, I have no idea how widespread this was: commonly attested Hebrew custom? General Mediterranean custom? Evidence?
So, without at the moment having evidence about evidence and Roman law, this connexion looks like a Just So story to me. The two words look alike: here is an odd verse in the Bible: and perhaps we might add the legend about Pope Joan and making sure subsequent popes had to have good dropped testicles: so here is a little story about the Romans. But did they really?
Personally I can't see why it shouldn't be just a slang use from POT. In English we give them good solid slang names, balls and nuts. Perhaps the old Romans called them pots. Note that if this is so, it would presumably have to have been before the later Romans began using testa POT to mean 'head'. Slang terms change.
Researching this, I came across another explanation in the American Heritage Dictionary. To me, as a linguist, this looks attractively plausible, because it doesn't require any stretching. It is a straightforward and normal example of how language works, and of what we know about technical borrowings between Latin and Greek. It is however dull: much less gripping than the picture of the Roman court the other theory conjures up.
In Greek, parastatês had two closely related meanings. It comes from para- 'by, beside' and *st(a)- 'stand': one who stands by or beside someone or something else. As a person it means a comrade beside you in battle, or an ally or a supporter in court, or witness. As a part of the body it means the testicles: the two things that stand by each other. (And related words mean things like door-posts.)
In Latin they had an ordinary word testis WITNESS. Then they wanted a medical term for GONAD so they borrowed from the Greek, as they often did. There are two possible ways of borrowing: take the actual word parastatês, or make a calque, i.e. borrow the idea that 'gonad' and 'witness' can be called by the same word. Note that this doesn't make sense on just the Latin words alone: there is no reason to connect gonad and witness as such; but it does make a lot of sense on the Greek breakdown of meaning para-st- 'by-stand-'. (In English we'd probably say lie or sit instead of stand, but it's just a general word for being located.)
So if testis was a medical borrowing, and its diminutive testiculus perhaps an invention of a slightly different word to make the distinction clear, what was the ordinary Roman in the street's word for balls or nuts? Looking at modern Romance languages, we notice Italian coglioni and Spanish cojones. If I go to my Latin dictionary I find a reasonable candidate for their origin: culleus 'leather bag for holding liquids'. That would have become colleo, collio in Vulgar Latin, so fits nicely with an augmentative collione.
W.W. Skeat, Concise Etymological Dictionary of the English Language
Shorter Oxford English Dictionary just says 'etym. dub.' (dubious) under testis GONAD
American Heritage Dictionary at http://www.bartleby.com/61/73/T0127300.html
Other dictionaries and sources are just repeating these ideas and should not be relied on.
Appendix on *TER-ST-I-S
Highly technical. Avoid if already confused by the above.
With thanks to drdave for raising possibilities. Overall I disagree, but herein is a way I've worked out by which the tre- 'three' etymology of WITNESS could be justified.
Now I've said *tre- becoming te- is unlikely, while *ters- become tes- is likely. So instead of the tre- forms of 'three', what about the ter- forms as in tertius 'third', and 'tertiary'? Could they be the answer?
Well the Proto-Indo-European root is *trey- with reduced grade *tri- (see ablaut for what grades mean, and in this case also sonant). It's not **ter-, reduced **tr-. The ordinal formative was -t-, which we also see in Greek tri-t-os, English thir-d, four-th. We would expect pre-Latin *tréi-t-os if it was accented on the first syllable, and reduced grade *tri-t-ós if it was accented on the -tós, as in fact it was, as evidence from other languages shows. Now the Latin ordinal ending was -os (later -us) as in quartus, quintus, sextus, and only tertius shows -ius. This looks very much like a metathesis *try-t-ós > *tr-t-yós. Then it developed an epenthetic vowel, giving the known form ter-t-ios. Then in Old Latin the accent shifted back onto the initial vowel. So that's how *trey- could give tertius.
Applying this to the proposed *trey-st-i-s, we have to be careful. We know the ordinal *-tós was accented, but now we want the i-stem thematic vowel to be also (and I would want to see examples of analogous formations before believing in this). If the -stís was accented then the syllable *trey- would drop to zero grade *tri-, giving *tri-st-í-s. This tristis is in fact the ordinary Classical Latin for 'sad', so there's nothing phonetically ungainly about it. No reason it should change.
But let's suppose it did metathesize by analogy with tertius. Then you'd possibly get the sequence of changes *tr-st-y-ís > *tr-st-ís > *terstís > *térstis > testis.