Chapter I: Part 1

Transcriber's Note: Phonetic characters are represented by the
following symbols:
[`x] = any letter "x" with grave accent
['x] = any letter "x" with acute accent
[:x] = any letter "x" with superior double-dot (dieresis)
[^x] = any letter "x" with superior circumflex
[=x] = any letter "x" with superior macron
[)x] = any letter "x" with superior breve
[e] = inverted "e" or schwa
[ae], [oe] = ae, oe ligature characters
[=xy] = any pair of letters "xy" with joining macron, except
[=OE], [=ae] = OE, ae ligature characters with macron and
['oe], ['ae] = oe, ae ligature characters with acute accent and
[)xy] = any pair of letters "xy" with joining breve, except
[)AE], [)ae], [)OE], [)oe] = AE, ae, OE, oe ligature characters
with breve
[^1] = raised "1", etc.

_S.P.E. TRACT NO. IV_

THE PRONUNCIATION OF ENGLISH WORDS DERIVED FROM THE LATIN

BY JOHN SARGEAUNT

WITH PREFACE AND NOTES BY H. BRADLEY

CORRESPONDENCE & MISCELLANEOUS NOTES BY H.B., R.B., W.H.F., AND EDITORIAL

_AT THE CLARENDON PRESS_ MDCCCCXX

ON THE PRONUNCIATION OF ENGLISH WORDS DERIVED FROM LATIN

[This paper may perhaps need a few words of introduction concerning the history of the pronunciation of Latin in England.

The Latin taught by Pope Gregory's missionaries to their English converts at the beginning of the seventh century was a living language. Its pronunciation, in the mouths of educated people when they spoke carefully, was still practically what it had been in the first century, with the following important exceptions. 1. The consonantal _u_ was sounded like the _v_ of modern English, 2. The _c_ before front vowels (_e_, _i_, _o_, _[ae]_, _[oe]_), and the combinations _t[)i]_, _c[)i]_ before vowels, were pronounced _ts_. 3. The _g_ before front vowels had a sound closely resembling that of the Latin consonantal _i_. 4. The _s_ between vowels was pronounced like our _s_. 5. The combinations _[ae]_, _[oe]_ were no longer pronounced as diphthongs, but like the simple _e_. 6. The ancient vowel-quantities were preserved only in the penultima of polysyllables (where they determined the stress); in all other positions the original system of quantities had given place to a new system based mainly on rhythm. Of this system in detail we have little certain knowledge; but one of its features was that the vowel which ended the first syllable of a disyllabic was always long: _p[=a]ter_, _p[=a]trem_, _D[=e]us_, _p[=i]us_, _[=i]ter_, _[=o]vis_, _h[=u]mus_.

Even so early as the beginning of the fifth century, St. Augustine tells us that the vowel-quantities, which it was necessary to learn in order to write verse correctly, were not observed in speech. The Latin-speaking schoolboy had to learn them in much the same fashion as did the English schoolboy of the nineteenth century.

It is interesting to observe that, while the English scholars of the tenth century pronounced their Latin in the manner which their ancestors had learned from the continental missionaries, the tradition of the ancient vowel-quantities still survived (to some extent at least) among their British neighbours, whose knowledge of Latin was an inheritance from the days of Roman rule. On this point the following passage from the preface to [AE]lfric's Latin Grammar (written for English schoolboys about A.D. 1000) is instructive:--

Miror ualde quare multi corripiunt sillabas in prosa quae in
metro breues sunt, cum prosa absoluta sit a lege metri; sicut
pronuntiant _pater_ brittonice et _malus_ et similia, quae in
metro habentur breues. Mihi tamen uidetur melius inuocare Deum
Patrem honorifice producta sillaba quam brittonice corripere,
quia nec Deus arti grammaticae subiciendus est.

The British contagion of which [AE]lfric here complains had no permanent effect. For after the Norman Conquest English boys learned their Latin from teachers whose ordinary language was French. For a time, they were not usually taught to write or read English, but only French and Latin; so that the Englishmen who attempted to write their native language did so in a phonetic orthography on a French basis. The higher classes in England, all through the thirteenth century, had two native languages, English and French.

In the grammar schools, the Latin lessons were given in French; it was not till the middle of the fourteenth century that a bold educational reformer, John Cornwall, could venture to make English the vehicle of instruction. In reading Latin, the rhythmically-determined vowel-quantities of post-classical times were used; and the Roman letters were pronounced, first as they were in French, and afterwards as in English, but in the fourteenth century this made little difference.

In Chaucer's time, the other nations of Europe, no less than England, pronounced Latin after the fashion of their own vernaculars. When, subsequently, the phonetic values of the letters in the vernacular gradually changed, the Latin pronunciation altered likewise. Hence, in the end, the pronunciation of Latin has become different in different countries. A scholar born in Italy has great difficulty in following a Frenchman speaking Latin. He has greater difficulty in understanding an Englishman's Latin, because in English the changes in the sounds of the letters have been greater than in any other language. Every vowel-letter has several sounds, and the normal long sound of every vowel-letter has no resemblance whatever to its normal short sound. As in England the pronunciation of Latin developed insensibly along with that of the native tongue, it eventually became so peculiar that by comparison the 'continental pronunciation' may be regarded as uniform.

It is sometimes imagined that the modern English way of pronouncing Latin was a deliberate invention of the Protestant reformers. For this view there is no foundation in fact. It may be conceded that English ecclesiastics and scholars who had frequent occasion to converse in Latin with Italians would learn to pronounce it in the Italian way; and no doubt the Reformation must have operated to arrest the growing tendency to the Italianization of English Latin. But there is no evidence that before the Reformation the un-English pronunciation was taught in the schools. The grammar-school pronunciation of the early nineteenth century was the lineal descendant of the grammar-school pronunciation of the fourteenth century.

This traditional system of pronunciation is now rapidly becoming obsolete, and for very good reasons. But it is the basis of the pronunciation of the many classical derivatives in English; and therefore it is highly important that we should understand precisely what it was before it began to be sophisticated (as in our own early days) by sporadic and inconsistent attempts to restore the classical quantities. In the following paper Mr. Sargeaunt describes, with a minuteness not before attempted, the genuine English tradition of Latin pronunciation, and points out its significance as a factor in the development of modern English.

H.B.]

* * * * *

It seems not to be generally known that there is a real principle in the English pronunciation of words borrowed from Latin and Greek, whether directly or through French. In this matter the very knowledge of classical Latin, of its stresses and its quantities, still more perhaps an acquaintance with Greek, is apt to mislead. Some speakers seem to think that their scholarship will be doubted unless they say 'doctr['i]nal' and 'script['u]ral' and 'cin['e]ma'. The object of this paper is to show by setting forth the principles consciously or unconsciously followed by our ancestors that such pronunciations are as erroneous as in the case of the ordinary man they are unnatural and pedantic. An exception for which there is a reason must of course be accepted, but an exception for which reason is unsound is on every ground to be deprecated. Among other motives for preserving the traditional pronunciation must be reckoned the claim of poetry. Mark Pattison notes how a passage of Pope which deals with the Barrier Treaty loses much of its effect because we no longer stress the second syllable of 'barrier'. Pope's word is gone beyond recovery, but others which are threatened by false theories may yet be preserved.

The _New English Dictionary_, whose business it is to record facts, shows that in not a few common words there is at present much confusion and uncertainty concerning the right pronunciation. This applies mostly to the position of the stress or, as some prefer to call it, the accent, but in many cases it is true also of the quantity of the vowels. It is desirable to show that there is a principle in this matter, rules which have been naturally and unconsciously obeyed, because they harmonize with the genius of the English tongue.

For nearly three centuries from the Reformation to the Victorian era there was in this country a uniform pronunciation of Latin. It had its own definite principles, involving in some cases a disregard of the classical quantities though not of the classical stress or accent. It survives in borrowed words such as _[=a]li[)a]s_ and _st[)a]mina_, in naturalized legal phrases, such as _N[=i]s[=i] Prius_ and _[=o]nus probandi_, and with some few changes in the Westminster Play. This pronunciation is now out of fashion, but, since its supersession does not justify a change in the pronunciation of words which have become part of our language, it will be well to begin with a formulation of its rules.

The rule of Latin stress was observed as it obtained in the time of Quintilian. In the earliest Latin the usage had been other, the stress coming as early in the word as was possible. Down to the days of Terence and probably somewhat later the old rule still held good of quadrisyllables with the scansion of _m[)u]l[)i][)e]r[)i]s_ or _m[)u]l[)i][)e]r[=e]s_, but in other words had given way to the later Quintilian rule, that all words with a long unit as penultimate had the stress on the vowel in that unit, while words of more than two syllables with a short penultimate had the stress on the antepenultimate. I say 'unit' because here, as in scansion, what counts is not the syllable, but the vowel plus all the consonants that come between it and the next vowel. Thus _inf['e]rnus_, where the penultimate vowel is short, no less than _supr['e]mus_, where it is long, has the stress on the penultima. In _volucris_, where the penultimate unit was short, as it was in prose and could be in verse, the stress was on the _o_, but when _ucr_ made a long unit the stress comes on the _u_, though of course the vowel remains short. In polysyllables there was a secondary stress on the alternate vowels. Ignorance of this usage has made a present-day critic falsely accuse Shakespeare of a false quantity in the line

Cor['i]ol['a]nus in Cor['i]oli.

It may be safely said that from the Reformation to the nineteenth century no Englishman pronounced the last word otherwise than I have written it. The author of the Pronouncing Dictionary attached to the 'Dictionary of Gardening' unfortunately instructs us to say _gl['a]diolus_ on the ground that the _i_ is short. The ground alleged, though true, is irrelevant, and, although Terence would have pronounced it _gl['a]diolus_, Quintilian, like Cicero, would have said _glad['i]olus_. Mr. Myles quotes Pliny for the word, but Pliny would no more have thought of saying _gl['a]diolus_ than we should now think of saying 'labo['u]r' except when we are reading Chaucer.

We need not here discuss the dubious exceptions to this rule, such as words with an enclitic attached, e.g. _prim[)a]que_ in which some authorities put the stress on the vowel which precedes the enclitic, or such clipt words as 'illuc', where the stress may at one time have fallen on the last vowel. In any case no English word is concerned.

In very long words the due alternation of stressed and unstressed vowels was not easy to maintain. There was no difficulty in such a combination as _h['o]nor['i]fic['a]bil['i]_ or as _tud['i]nit['a]tib['u]s_, but with the halves put together there would be a tendency to say _h['o]nor['i]ficabilit['u]dinit['a]tibus_. Thus there ought not to be much difficulty in saying _C['o]nstant['i]nop['o]lit['a]ni_, whether you keep the long antepenultima or shorten it after the English way; but he who forced the reluctant word to end an hexameter must have had 'Constantin['o]ple' in his mind, and therefore said _Const['a]ntin['o]polit['a]ni_ with two false stresses. The result was an illicit lengthening of the second _o_. His other false quantity, the shortening of the second _i_, was due to the English pronunciation, the influence of such words as 'metropol[)i]tan', and, as old schoolmasters used to put it, a neglect of the Gradus. Even when the stress falls on this antepenultimate _i_, it is short in English speech. Doubtless Milton shortened it in 'Areopagitica', just as English usage made him lengthen the initial vowel of the word.

Probably very few of the Englishmen who used the traditional pronunciation of Latin knew that they gave many different sounds to each of the symbols or letters. Words which have been transported bodily into English will provide examples under each head. It will be understood that in the traditional pronunciation of Latin these words were spoken exactly as they are spoken in the English of the present day. For the sake of simplicity it may be allowed us to ignore some distinctions rightly made by phoneticians. Thus the long initial vowel of _alias_ is not really the same as the long initial vowel of _area_, but the two will be treated as identical. It will thus be possible to write of only three kinds of vowels, long, short, and obscure.

The letter or symbol _a_ stood for two long sounds, heard in the first syllables of _alias_ and of _larva_, for the short sound heard in the first syllable of _stamina_, and for the obscure sound heard in the last syllable of each of these last two words in English.

The letter _e_ stood for the long sounds heard in _genus_ and in _verbum_, for the short sound heard in _item_, and for the obscure sound heard in _cancer_. When it ended a word it had, if short, the sound of a short _i_, as in _pro lege_, _rege_, _grege_, as also in unstressed syllables in such words as _precentor_ and _regalia_.

The letter _i_ stood for the two long sounds heard in _minor_ and in _circus_ and for the short sound heard in _premium_ and _incubus_.

The letter _o_ stood for the two long sounds heard in _odium_ and in _corpus_, for the short sound in _scrofula_, and for the obscure in _extempore_.

The two long sounds of _u_ are heard in _rumor_, if that spelling may be allowed, and in the middle syllable of _laburnum_, the two short sounds in the first _u_ of _incubus_ and in the first _u_ of _lustrum_, the obscure sound in the final syllables of these two words. Further the long sound was preceded except after _l_ and _r_ by a parasitic _y_ as in _albumen_ and _incubus_. This parasitic _y_ is perhaps not of very long standing. In some old families the tradition still compels such pronunciations as _moosic_.

The diphthongs _[ae]_ and _[oe]_ were merely _e_, while _au_ and _eu_ were sounded as in our _August_ and _Euxine_. The two latter diphthongs stood alone in never being shortened even when they were unstressed and followed by two consonants. Thus men said _[=Eu]stolia_ and _[=Au]gustus_, while they said _[)[AE]]schylus_ and _[)OE]dipus._ Dryden and many others usually wrote the _[AE]_ as _E_. Thus Garrick in a letter commends an adaptation of 'Eschylus', and although Boswell reports him as asking Harris 'Pray, Sir, have you read Potter's _[AE]schylus_?' both the speaker and the reporter called the name _Eschylus_.

The letter _y_ was treated as _i_.

The consonants were pronounced as in English words derived from Latin. Thus _c_ before _e_, _i_, _y_, _[ae]_, and _[oe]_ was _s_, as in _census_, _circus_, _Cyrus_, _C[ae]sar_, and _c[oe]lestial_, a spelling not classical and now out of use. Elsewhere _c_ was _k_. Before the same vowels _g_ was _j_ (d[ezh]), as in _genus_, _gibbus_, _gyrus_. The sibilant was voiced or voiceless as in English words, the one in _rosaceus_, the other in _saliva_.

It will be seen that the Latin sounds were throughout frankly Anglicized. According to Burney a like principle was followed by Burke when he read French poetry aloud. He read it as though it were English. Thus on his lips the French word _comment_ was pronounced as the English word _comment_.

The rule that overrode all others, though it has the exceptions given below, was that vowels and any other diphthongs than _au_ and _eu_, if they were followed by two consonants, were pronounced short. Thus _a_ in _magnus_, though long in classical Latin, was pronounced as in our 'magnitude', and _e_ in _census_, in Greek transcription represented by [Greek: eta], was pronounced short, as it is when borrowed into English. So were the penultimate vowels in _villa_, _nullus_, _c[ae]spes_.

This rule of shortening the vowel before two consonants held good even when in fact only one was pronounced, as in _nullus_ and other words where a double consonant was written and in Italian pronounced.

Moreover, the parasitic _y_ was treated as a consonant, hence our 'v[)a]cuum'.

In the penultima _qu_ was treated as a single consonant, so that the vowel was pronounced long in _[=a]quam_, _[=e]quam_, _in[=i]quam_, _l[=o]quor_. So it was after _o_, hence our 'coll[=o]quial'; but in earlier syllables than the penultima _qu_ was treated as a double consonant, hence our 'sub[)a]queous', 'equity', 'iniquity'.

EXCEPTIONS.

1. When the former of the two consonants was _r_ and the latter another consonant than _r_, as in the series represented by _larva_, _verbum_, _circus_, _corpus_, _laburnum_, the vowels are a separate class of long vowels, though not really recognized as such. Of course our ancestors and the Gradus marked them long because in verse the vowel with the two consonants makes a long unit.

2. A fully stressed vowel before a mute and _r_, or before _d_ or _pl_, was pronounced long in the penultima. Latin examples are _labrum_, _Hebrum_, _librum_, _probrum_, _rubrum_, _acrem_, _cedrum_, _vafrum_, _agrum_, _pigrum_, _aprum_, _veprem_, _patrem_, _citrum_, _utrum_, _triplus_, _duplex_, _Cyclops_. Moreover, in other syllables than the penultima the vowel in the same combinations was pronounced long if the two following vowels had no consonant between them, as _patria_, _Hadria_, _acrius_. (Our 'triple' comes from _triplum_ and is a duplicate of '_treble_'. Perhaps the short vowel is due to its passage through French. Our 'citron' comes from _citronem_, in which _i_ was short.)

3. The preposition and adverb _post_ was pronounced with a long vowel both by itself and in composition with verbs, but its adjectives did not follow suit. Hence we say in English 'p[=o]stpone', but 'p[)o]sterior' and 'p[)o]sthumous'.

Monosyllables ending in a vowel were pronounced long, those ending in a consonant short. Enclitics like _que_ were no real exception as they formed part of the preceding word. There were, however, some real exceptions.

1. Pronouns ending in _-os_, as _hos_, _quos_. These followed _eos_ and _illos_.

2. Words ending in _-es_, as _pes_, _res_.

3. Words ending in _r_, as _par_, _fer_, _vir_, _cor_, _fur_. These had that form of long vowel which we use in 'part', 'fertile', 'virtue', 'cordate', 'furtive'.

In, disyllables the former vowel or diphthong, if followed by a single consonant, or by a mute and _r_, or by _cl_ or _pl_, was pronounced long, a usage which according to Mr. Henry Bradley dates in spoken Latin from the fourth century. Examples are _apex_, _tenet_, _item_, _focus_, _pupa_, _Psyche_, _C[ae]sar_, _f[oe]tus_. I believe that at first the only exceptions were _tibi_, _sibi_, _ibi_, _quibus_, _tribus_. In later days the imperfect and future of _sum_ became exceptions. Here perhaps the short vowel arose from the hideous and wholly erroneous habit, happily never universal though still in some vogue, of reciting _er['a]m_, _er['a]s_, _er['a]t_. There are actually schoolbooks which treat the verse _ictus_, the beat of the chanter's foot, as a word stress and prescribe _terra trib['u]s scopul['i]s_. I can say of these books only _Pereant ipsi, mutescant scriptores_, and do not mind using a post-classical word in order to say it.

In disyllables the former vowel or diphthong, if followed immediately by another vowel or diphthong, had the quality, and if emphatic also the quality, of a long vowel. The distinction was not recognized, and seems not to be generally acknowledged even now. We seem not to have borrowed many words which will illustrate this. We have however _fiat_, and _pius_ was pronounced exactly as we pronounce 'pious', while for a diphthong we may quote Shelley,

Mid the mountains Euganean
I stood listening to the paean.

English derivatives will show the long quality of the vowels in _aer_, _deus_, _coit_, _duo_. To these add _Graius_.

The rule of _apex_ applies also to words of more than two syllables with long penultima, as _gravamen_, _arena_, _saliva_, _abdomen_, _acumen_. The rule of _aer_ also holds good though it hardly has other instances than Greek names, as _Mach['a]on_, _[AE]n['e]as_, _Thal['i]a_, _Achel['o]us_, _Ach['[ae]]i_.

In words of more than two syllables with short penultima the vowel in the stressed antepenultima was pronounced short when there was a consonant between the two last vowels, and _i_ and _y_ were short even when no consonant stood in that place. Examples are _stamina_, _Sexagesima_, _minimum_, _modicum_, _tibia_, _Polybius_. But _u_, _au_, _eu_ were, as usual, exceptions, as _tumulus_, _Aufidus_, _Eutychus_. I believe that originally men said _C[)[ae]]sarem_, as they certainly said _c[)[ae]]spitem_ and _C[)[ae]]tulum_, as also _C[)[ae]]sarea_, but here in familiar words the cases came to follow the nominative.

Exceptions to the rule were verb forms which had _[=a]v_, _[=e]v_, _[=i]v_, or _[=o]v_ in the antepenultima, as _am[=a]veram_, _defieverat_, _audivero_, _moveras_, and like forms from aorists with the penultima long, as _suaseram_, _egero_, _miserat_, _roseras_, and their compounds.

This rule was among the first to break down, and about the middle of the nineteenth century the Westminster Play began to observe the true quantities in the antepenultimate syllables. Thus in spite of 'cons[)i]deration' boys said _s[=i]dera_, and in spite of 'n[)o]minal' they said _n[^o]mina_, while they still said _s[)o]litus_ and _r[)a]pidus_.

On the other hand the following rule, of which borrowed words provide many examples, still obtains in the Play. In words of more than two syllables any vowel in the antepenultima other than _i_ or _y_ was pronounced long if no consonant divided the two following vowels. Possibly the reason was that there was a syn[ae]resis of the two vowels, but I doubt this, for a parasitic _y_ was treated as a consonant. Examples are _alias_, _genius_, _odium_, _junior_, _an[ae]mia_, and on the other hand _f[)i]lius_, _L[)y]dia_. Compound verbs with a short prefix were exceptions, as _[)o]beo_, _r[)e]creo_, whence our 'recreant'. A long prefix remained long as in _d[=e]sino_. The only other exception that I can remember was _Ph[)o]loe_.

In polysyllables the general rule was that all vowels and diphthongs before the penultima other than _u_, when it bore a primary or secondary stress, and _au_ and _eu_ were pronounced short except where the 'alias' rule or the 'larva' rule applied. Thus we said _h[)e]r[)e]ditaritis_, _[)[ae]]qu[)a]bilitas_, _imb[)e]cillus_, _susp[)i]cionem_, but _fid[=u]ciarius_, _m[=e]diocritas_, _p[=a]rticipare_. I do not know why the popular voice now gives _[)A]riadne_, for our forefathers said _[=A]riadne_ as they said _[=a]rea_.

In very long words the alternation of stress and no-stress was insisted on. I remember a schoolmaster who took his degree at Oxford in the year 1827 reproving a boy for saying _['A]lphesib['oe]us_ instead of _Alphesib['oe]us_, and I suspect that Wordsworth meant no inverted stress in

La['o]dam['i]a, that at Jove's command--

nor Landor in

Art['e]mid['o]ra, gods invisible--

though I hope that they did.

* * * * *

It is not to be thought that these rules were in any way arbitrary. So little was this so that, I believe, they were never even formulated. If examples with the quantities marked were ever given, they must have been for the use of foreigners settling in England. English boys did not want rules, and their teachers could not really have given them. The teachers did not understand that each vowel represented not two sounds only, a long and a short, but many more. This fact was no more understood by John Walker, the actor and lexicographer, who in 1798 published a Key to the Classical Pronunciation of Greek and Latin proper names. His general rule was wrong as a general rule, and so far as it agreed with facts it was useless. He says that when a vowel ends a syllable it is long, and when it does not it is short. Apart from the confusion of cause and effect there is the error of identifying for instance the _e_ in _beatus_ and the _e_ in _habebat_. Moreover, Walker confounds the _u_ in 'curfew', really long, with the short and otherwise different _u_ in 'but'. The rule was useless as a guide, for it did not say whether _moneo_ for instance was to be read as _ino-neo_ or as _mon-eo_, and therefore whether the _o_ was to be long or short. Even Walker's list is no exact guide. He gives for instance _M[=o]-na_, which is right, and _M[=o]-n[ae]ses_, which is not. Now without going into the difference between long vowels and ordinary vowels, of which latter some are long in scansion and some short, it is clear that there is no identity. In fact _Mona_, has the long _o_ of 'moan' and _Mon[ae]ses_ the ordinary _o_ of 'monaster'. A boy at school was not troubled by these matters. He had only two things to learn, first the quantity of the penultimate unit, second the fact that a final vowel was pronounced. When he knew these two things he gave the Latin word the sounds which it would have if it were an English word imported from the Latin. Thus he finds the word _civilitate_. I am not sure that he could find it, but that does not matter. He would know 'civility', and he learns that the penultima of the Latin word is long. Therefore he says _c[)i]v[)i]l[)i]t[=a]t[)e]_. Again he knows '[)i]nf[)i]n[)i]t' (I must be allowed to spell the word as it is pronounced except in corrupt quires). He finds that the penultima of _infinitivus_ is long, and he therefore says _[)i]nf[)i]n[)i]t[=i]v[)u]s_. Again he knows 'irradiate', and finding that the penultima of _irradiabitur_ is short he says _[)i]rr[=a]d[)i][)a]b[)i]t[)u]r_. It is true that some of these verb forms under the influence of their congeners came to have an exceptional pronunciation. Thus _irradi[=a]bit_ led at last to _irradi[=a]bitur_, but I doubt whether this occurred before the nineteenth century. The word _dabitur_, almost naturalized by Luther's adage of _date et dabitur_, kept its short _a_ down to the time when it regained it, in a slightly different form, by its Roman right; and _am[)a]mini_ and _mon[)e]mini_ were unwavering in their use. Old people said _v[=a]ri[)a]bilis_ long after the true quantities had asserted themselves, and the word as the specific name of a plant may be heard even now. Its first syllable of course follows what I shall call the 'alias' rule. We may still see this rule in other instances. All men say 'hippop['o]t[)a]mus', and even those who know that this _a_ is short in Greek can say nothing but 'Mesopot[=a]mia', unless indeed the word lose its blessed and comforting powers in a disyllabic abbreviation. When a country was named after Cecil Rhodes, where the _e_ in the surname is mute, we all called it 'Rhod[=e]sia'. Had it been named after a Newman, where the _a_ is short or rather obscure, we should all have called it 'Newm[=a]nia ', while, named after a Davis, it would certainly have been 'Dav[)i]sia'. The process of thought would in each case have been unconscious. A new example is 'aviation', whose first vowel has been instinctively lengthened.

Again, when the word 'telegram' was coined, some scholars objected to its formation and insisted upon 'telegrapheme', but the most obdurate Grecian did not propose to keep the long Greek vowel in the first syllable. When only the other day 'cinematograph' made its not wholly desirable appearance, it made no claim to a long vowel in either of its two first syllables. Not till it was reasonably shortened into 'c[)i]n[)e]ma' did a Judge from the Bench make a lawless decree for a long second vowel, and even he left the _i_ short though it is long in Greek.

Of course with the manner of speech the quantities had to be learnt separately. The task was not as difficult as some may think. To boys with a taste for making verses the thumbing of a Gradus (I hope that no one calls it a Gr[)a]dus) was always a delightful occupation, and a quantity once learnt was seldom forgotten. It must be admitted that, as boys were forced to do verses, whether they could or not, there were always some who could read and yet forget.

Although these usages did not precede but followed the pronunciation of words already borrowed from Latin, we may use them to classify the changes of quantity. We shall see that although there are some exceptions for which it is difficult to give a reason, yet most of the exceptions fall under two classes. When words came to us through French, the pronunciation was often affected by the French form of the word. Thus the adjective 'present' would, if it had come direct from Latin, have had a long vowel in the first syllable. To an English ear 'pr[)e]sent' seemed nearer than 'pr[=e]sent' to the French 'pr['e]sent'. The _N.E.D._ says that 'gladiator' comes straight from the Latin 'gladiatorem'. Surely in that case it would have had its first vowel long, as in 'radiator' and 'mediator'. In any case its pronunciation must have been affected by 'gladiateur'. The other class of exceptions consists of words deliberately introduced by writers at a late period. Thus 'adorable' began as a penman's word. Following 'in['e]xorable' and the like it should have been '['a]dorable'. Actually it was formed by adding _-able_ to 'ad['o]re', like 'laughable'. It is now too stiff in the joints to think of a change, and must continue to figure with the other sins of the Restoration.

Before dealing with the words as classified by their formation, we may make short lists of typical words to show that for the pronunciation of English derivatives it is idle to refer to the classical quantities.

From _[=[ae]]_: [)e]difice, [)e]mulate, c[)e]rulean, qu[)e]stion.

From _[=oe]_: [)e]conomy, [)e]cumenical, conf[)e]derate.

From _[=a]_,: don[)a]tive, n[)a]tural, cl[)a]mour, [)a]verse.

From _[)a]_: [=a]lien, st[=a]tion, st[=a]ble, [=a]miable.

From _[=e]_: [)e]vident, Quadrag[)e]sima, pl[)e]nitude, s[)e]gregate.

From _[)e]_: s[=e]ries, s[=e]nile, g[=e]nus, g[=e]nius.

From _[=i]_: lasc[)i]vious, erad[)i]cate, d[)i]vidend, f[)i]lial, susp[)i]cion.

From _[)i]_: l[=i]bel, m[=i]tre, s[=i]lex.

From _[=o]_: [)o]rator, pr[)o]minent, pr[)o]montory, s[)o]litude.

From _[)o]_: b[=o]vine, l[=o]cal, f[=o]rum, coll[=o]quial.

From _[=u]_: fig[)u]rative, script[)u]ral, sol[)u]ble.

From _[)u]_: n[=u]merous, C[=u]pid, all[=u]vial, cer[=u]lean.

The _N.E.D._ prefers the spelling '[oe]cumenical'; but Newman wrote naturally 'ecumenical', and so does Dr. J.B. Bury. Dublin scholarship has in this matter been markedly correct.

_CLASSIFICATION OF WORDS ACCORDING TO THEIR LATIN STEMS._

In classification it seems simplest to take the words according to their Latin stems. We must, however, first deal with a class of adjectives borrowed bodily from the Latin nominative masculine with the insertion of a meaningless _o_ before the final _-us_.[1] These of course follow the rules given above. In words of more than two syllables the antepenultimate and stressed vowel is shortened, as '[)e]mulous' from _[ae]mulus_ and in 'fr[)i]volous' from _fr[=i]volus_, except where by the 'alias' rule it is long, as in 'egr[=e]gious' from _egr[)e]gius_. Words coined on this analogy also follow the rules. Thus 'glabrous' and 'fibrous' have the vowels long, as in the traditional pronunciation of _glabrum_ and _fibrum_, where the vowels in classical Latin were short. The stressed _u_ being always long we have 'lug[=u]brious' and 'sal[=u]brious', the length being independent of the 'alias' rule. Some words ending in _-ous_ are not of this class. Thus 'odorous' and 'clamorous' appear in Italian as _odoroso_ and _clamoroso_. Milton has

Son['o]rous mettal blowing Martial sounds.

The Italian is _sonoro_, and our word was simply the Latin _sonorus_ borrowed bodily at a somewhat late period. Hence the stress remains on the penultima. Skeat thought that the word would at last become 's['o]norous'. It maybe hoped that Milton's line will save it from the effect of a false analogy.

[Footnote 1: I regard this statement as inaccurate. The _-ous_ in these words does not come from the nominative ending _-us_, but is the ordinary _-ous_ from L. _-osus_ (through Fr.). It was added to many Latin adjective stems, because the need of a distinctly adjectival ending was felt. Similarly in early French _-eux_ was appended to adjectives when they were felt to require a termination, as in _pieux_ from _pi-us_. Compare the English _capacious_, _veracious_, _hilarious_, where _-ous_ is added to other stems than those in _o_. Other suffixes of Latin origin are used in the same way: e.g. _-al_ in _aerial_, _ethereal_.--H.B.]

In classifying by stems it will be well to add, where possible, words of Greek origin. Except in some late introductions Greek words, except when introduced bodily, have been treated as if they came through Latin, and some of the bodily introductions are in the same case. Thus 'an[ae]sthetic' is spelt with the Latin diphthong and the Latin _c_. Even 'skeleton' had a _c_ to start with, while the modern and wholly abominable 'kaleidoscope' is unprincipled on the face of it.

STEMS ENDING IN -ANT AND -ENT. These are participles or words formed as such. Our words have shed a syllable, thus _regentem_ has become 'regent'. Disyllables follow the 'apex' rule and lengthen the first vowel, as 'agent', 'decent', 'potent'. Exceptions are 'clement' and 'present', perhaps under French influence. Words of more than two syllables with a single consonant before the termination throw the stress back and shorten a long penultima, as 'ignorant', 'president', 'confident', 'adjutant'. Where there are two heavy consonants, the stress remains on the penultima, as 'consultant', 'triumphant', even when one of the consonants is not pronounced, as 'reminiscent'. In some cases the Latinists seem to have deliberately altered the natural pronunciation. Thus Gower has '['a]ppara['u]nt', but the word became 'app['a]rent' before Shakespeare's time, and later introductions such as 'adherent' followed it. What right 'adjacent' has to its long vowel and penultimate stress I do not know, but it cannot be altered now.

STEMS ENDING IN -ATO AND -UTO. These are mostly past participles, but many of them are used in English as verbs. It must be admitted that the disyllabic words are not wholly constant to a principle. Those verbs that come from _-latum_ consistently stress the last vowel, as 'dilate', 'relate', 'collate'. So does 'create', because of one vowel following another. Of the rest all the words of any rank have the stress on the penultima, as 'vibrate', 'frustrate', 'm['i]grate', 'c['a]strate', 'p['u]lsate', 'v['a]cate'. Thus Pope has

The whisper, that to greatness still too near,
Perhaps, yet vibrates on his Sov'reign's ear,

and Shelley

Music, when soft voices die,
Vibrates in the memory.

There are, however, verbs of no literary account which in usage either vary in the stress or take it on the latter syllable. Such are 'locate', 'orate', 'negate', 'placate', and perhaps 'rotate'. With most of these we could well dispense. 'Equate' is mainly a technical word. Dictionaries seem to prefer the stress on the ultima, but some at least of the early Victorian mathematicians said '['e]quate', and the pronunciation is to be supported. Trisyllabic verbs throw the stress back and shorten the penultima, as 'd['e]s[)o]late', 's['u]ff[)o]cate', 'sc['i]nt[)i]llate'. Even words with heavy double consonants have adopted this habit. Thus where Browning has (like Milton and Cowper)

I the Trinity ill['u]strate
Drinking water'd orange pulp,
In three sips the Arian frustrate.
While he drains his at one gulp,

it is now usual to say '['i]llustrate'.

Adjectives of this class take as early a stress as they can, as '['o]rnate', 'p['i]nnate', 'd['e]licate', 'f['o]rtunate'. Nouns from all these words throw the accent back and shorten or obscure all but the penultimate vowel, as 'ignorance', 'evaporation'.

STEMS IN -IA. Here even disyllables shorten the penultima, as 'copy', 'province', while longer words throw the stress back as well as shorten the penultima, as 'injury', 'colony', while 'ignominy' almost lost its penultimate vowel, and therefore threw back the stress to the first syllable. Shakespeare frankly writes the word as a trisyllable,

Thy ignomy sleep with thee in the grave.

Milton restored the lost syllable, often eliding the final vowel, as in

Exile, or ignominy, or bonds, or pain.

Even with heavy consonants we have the early stress, as in 'industry'. Greek words follow the same rules, as 'agony', 'melody'. Some words of this class have under French influence been further abbreviated, as 'concord'.

Corresponding STEMS IN -IO keep the same rules. Perhaps the only disyllable is 'study'; the shortening of a stressed _u_ shows its immediate derivation from the old French _estudie_. Trisyllabic examples are 'colloquy', 'ministry', 'perjury'. Many words of this class have been further abbreviated in their passage through French. Such are 'benefice', 'divorce', 'office', 'presage', 'suffrage', 'vestige', 'adverb', 'homicide', 'proverb'. The stress in 'div['o]rce' is due to the long vowel and the two consonants. A few of these words have been borrowed bodily from Latin, as 'odium', 'tedium', 'opprobrium'.

STEMS IN -DO AND -TO (-SO). These words lose the final Latin syllable and keep the stress on the vowel which bore it in Latin. The stressed vowel, except in _au_, _eu_, is short, even when, as in 'vivid', 'florid', it was long in classical Latin. This, of course, is in accord with the English pronunciation of Latin. Examples are 'acid', 'tepid', 'rigid', 'horrid', 'humid', 'lurid ', 'absurd', 'tacit', 'digit', 'deposit', 'compact', 'complex', 'revise', 'response', 'acute'. Those which have the suffix _-es_ prefixed throw the stress back, as 'honest', 'modest'. Those which have the suffix _-men_ prefixed also throw the stress back, as 'moment', 'pigment', 'torment', and to the antepenultima, if there be one, as 'argument', 'armament', 'emolument', the penultimate vowel becoming short or obscure. In 'temperament' the tendency of the second syllable to disappear has carried the stress still further back. We may compare 'S['e]ptuagint', where _u_ becomes consonantal. An exception for which I cannot account is 'cem['e]nt', but Shakespeare has 'c['e]ment'.

STEMS IN -T[=A]T. These are nouns and have the stress on the antepenultima, which in Latin bore the secondary stress. They of course show the usual shortening of the vowels with the usual exceptions. Examples are 'charity', 'equity', 'liberty', 'ferocity', 'authority', and with long antepenultima 'immunity', 'security', 'university'. With no vowel before the penultima the long quality is, as usual, preserved, as in 'satiety'.

STEMS IN -OSO. These are adjectives and throw the stress back to the antepenultima, if there be one. In disyllables the penultimate vowel is long, as in 'famous', 'vinous'; in longer words the antepenultimate vowel is short, as 'criminous', 'generous'. Many, however, fall under the 'alias' rule, as 'ingenious', 'odious', while those which have _i_ in the penultimate run the two last syllables into one, as 'pernicious', 'religious', 'vicious'. A few late introductions, coming straight from the Latin, retained the Latin stress, as 'morose', 'verbose'.

STEMS IN -T[=O]RIO AND -S[=O]RIO. In these words the stress goes back to the fourth syllable from the end, this in Latin having the secondary stress, or, as in 'circulatory', 'ambulatory', even further. In fact the _o_, which of course is shortened, tends to disappear. Examples are 'declamatory', 'desultory', 'oratory', 'predatory', 'territory'. Three consonants running, as in 'perfunctory', keep the stress where it has to be in a trisyllable, such as 'victory'. So does a long vowel before _r_ and another consonant, as in 'precursory'. Otherwise two consonants have not this effect, as in 'pr['o]montory', 'c['o]nsistory'. In spite of Milton's

A gloomy Consistory, and them amidst
With looks agast and sad he thus bespake,

the word is sometimes mispronounced.

STEMS IN -[=A]RIO. These follow the same rules, except that, as in '['a]dversary', combinations like _ers_ are shortened and the stress goes back; and that words ending in _-entary_, such as 'elementary' and 'testamentary', stress the antepenultima. Examples are 'antiquary', 'honorary', 'voluntary', 'emissary'. It is difficult to see a reason for an irregular quantity in the antepenultima of some trisyllables. The general rule makes it short, as in 'granary', 'salary', but in 'library' and 'notary' it has been lengthened. The _N.E.D._ gives 'pl[=e]nary', but our grandfathers said 'pl[)e]nary'. Of course 'diary' gives a long quality to the _i_.

STEMS IN -[)I]LI. These seem originally to have retained the short _i_. Thus Milton's spelling is 'facil' and 'fertil' while other seventeenth-century writers give 'steril'. This pronunciation still obtains in America, but in England the words seem to have been usually assimilated to 'fragile', as Milton spells it, which perhaps always lengthened the vowel. The penultimate vowel is short.

STEMS IN -[=I]LI. Here the long _i_ is retained, and in disyllables the penultima is lengthened, as in 'anile', 'senile', 'virile'. There is no excuse for following the classical quantity in the former syllables of any of these words. As an English word 'sedilia' shortens the antepenultimate, like 'tibia' and the rest, the 'alias' rule not applying when the vowel is _i_.

STEMS IN -B[)I]LI. These mostly come through French and change the suffix into _-ble_. Disyllables lengthen the penultima, as 'able', 'stable', 'noble', while 'mobile', as in French, lengthens its latter vowel. Trisyllables shorten and stress the antepenultima, as 'placable', 'equable', but of course _u_ remains long, as in 'mutable'. Longer words throw the stress further back, except mere negatives, like 'impl['a]cable', and words with heavy consonants such as 'delectable'. Examples are 'miserable', 'admirable', 'intolerable', 'despicable'. The Poet Laureate holds that in these words Milton kept the long Italian _a_ of the penultimate or secondary stress.

Fall'n Cherube, to be weak is miserable.

In English we have naturalized _-able_ as a suffix and added it to almost any verb, as 'laughable', 'indescribable', 'desirable'. The last word may have been taken from French. The form 'des[)i]derable' occurs from the fourteenth to the seventeenth century. Originally 'acceptable' threw the stress back, as in Milton's

So fit, so acceptable, so Divine,

but the double mute has brought it into line with 'delectable'. Nowadays one sometimes hears 'disp['u]table', 'desp['i]cable', but these are intolerable vulgarisms.

SUFFIXES IN T[)I]LI AND S[)I]LI. These words mostly lengthen the _i_ and make the usual shortenings, as 'missile', 'sessile', 'textile', 'volatile', but of course 'futile'. Exceptions which I cannot explain are 'foss[)i]l' and 'fus[)i]le'.

SUFFIX IN [=A]LI. These adjectives shorten the _-a_ and, with the usual exceptions, the preceding vowels, as 'd['o]ctrinal', 'f['i]lial', 'l['i]beral', 'm['a]rital', 'med['i]cinal', but of course by the 'alias' rule 'arb[=o]real' (not a classical word in Latin) and 'g[=e]nial'. Words like 'national' and 'rational' were treated like trisyllables, which they now are. The stress is on the antepenultima except when heavy consonants bring it on to the penultima, as in 'sepulcral', 'parental', 'triumphal'.

Those who say 'doctr['i]nal' on the ground that the second vowel is long in Latin commit themselves to 'medic['i]nal', 'nat['u]ral', 'nutr['i]ment', 'instr['u]ment', and, if their own principle be applied, they make false quantities by the dozen every day of their lives.

Three words mostly mispronounced are, from their rarity, perhaps not past rescue. They are 'd['e]canal', 'rurid['e]canal', and 'pr['e]bendal'. There is no more reason for saying 'dec['a]nal' than for saying 'matr['o]nal' or for saying 'preb['e]ndal' than for saying 'cal['e]ndar'. Of course words like 'tremendous', being imported whole, keep the original stress. In our case the Latin words came into existence as _d['e]can['a]lis_, _pr['e]bend['a]lis_, parallel with _n['a]tur['a]lis_, which gives us 'n['a]tural'. That mostly wrong-headed man, Burgon of Chichester, was correct in speaking of his rights or at any rate his claims as 'd['e]canal'.

STEMS IN -LO. Of these 'stimulus' and 'villa' have been borrowed whole, while _umbella_ is corrupted into 'umbrella'. Disyllables lengthen the penultima, as 'stable', 'title', 'pupil'. Under French influence 'disciple' follows their example. In longer words the usual shortenings are made, as in 'frivolous', 'ridiculous'. The older words in _-ulo_ change the suffix into _-le_, as 'uncle', 'maniple', 'tabernacle', 'conventicle', 'receptacle', 'panicle'. Later words retain the _u_, as 'vestibule', 'reticule', 'molecule'.

STEMS IN -NO. The many words of this class are a grief to the classifier, who seeks in vain for reasons. Thus 'german' and 'germane' have the same source and travelled, it seems, by the same road through France. The Latin _hyacinth[)i]nus_ and _adamant[)i]nus_ are parallel words, yet Milton has 'hyacinthin' for the one and 'adamantine' for the other. One classification goes a little way. Thus 'human' and 'urban' must have come through French, 'humane' and 'urbane' direct from Latin. On the other hand while 'meridian' and 'quartan' are French, 'publican', 'veteran', and 'oppidan' are Latin. Words with a long _i_, if they came early through France, shorten the vowel, as 'doctrine', 'discipline', 'medicine', and 'masculine', while 'genuine', though a later word, followed them, but 'anserine' and 'leonine' did not. Disyllables seem to prefer the stress on the ultima, as 'divine', 'supine', but even these are not consistent. Some critics would scan Cassio's words

The d['i]vine Desdemona,

though Shakespeare nowhere else has this stress, while Shelley has. Shelley, too, has

She cannot know how well the s['u]pine slaves
Of blind authority read the truth of things.

The grammatical term, too, is 's['u]pine'. Later introductions also have this stress, as 'b['o]vine', 'c['a]nine', '['e]quine'. The last word is not always understood. At any rate Halliwell-Phillips, referring to a well-known story of Shakespeare's youth, says that the poet probably attended the theatre 'in some equine capacity'. As it is agreed that 'bovine' and 'equine' lengthen the former vowel, we ought by analogy to say 'c[=a]nine', as probably most people do. Words of more than two syllables have the stress on the antepenultima and the vowel is short, as in 'libertine', 'adulterine', but of course '[=u]terine'. When heavy consonants bring the stress on to the penultima, the _i_ is shortened, as in 'clandest[)i]n(e)', 'intest[)i]n(e)', and so in like disyllables, as 'doctr[)i]n(e)'. The modern words 'morphin(e)' and 'strychnin(e)', coined, the one from Morpheus and the other from the Greek name of the plant known to botanists as _Withania somnifera_, correctly follow 'doctrine' in shortening the _i_, though another pronunciation is sometimes heard.

STEMS IN -TUDIN. These shorten the antepenultima, as 'plenitude', 'solitude', with the usual exceptions, such as 'fortitude'.

STEMS IN -TION. These words retain the suffix, which in early days was disyllabic, as it sometimes is in Shakespeare, for instance in Portia's

Before a friend of this descripti['o]n
Shall lose a hair through Bassanio's fault.

Thus they came under the 'alias' rule, and what is now the penultimate vowel is long unless it be _i_. Examples are 'nation', 'accretion', 'emotion', 'solution', while _i_ is shortened in 'petition', 'munition', and the like, and left short in 'admonition' and others. In military use an exception is made by 'ration', but the pronunciation is confined to one sense of the word, and is new at that. I remember old soldiers of George III who spoke of 'r[=a]tions'. Perhaps the ugly change is due to French influence.

Originally the adjectives from these words must have lengthened the fourth vowel from the end long, as n[=a]t[)i][)o]nal, but when _ti_ became _sh_ they came to follow the rule of Latin trisyllables in our pronunciation.

STEMS IN -IC. Of these words we have a good many, both Latin and Greek. Those that came direct keep the stress on the vowel which was antepenultimate and is in English penultimate, and this vowel is short whatever its original quantity. Examples are 'aquatic', 'italic', 'Germanic'. Words that came through French threw the stress back, as 'l['u]natic'. Skeat says that 'fanatic' came through French, but he can hardly be right, for the pronunciation 'f['a]natic' is barely three score years old. There is no inverted stress in Milton's

Fan['a]tic Egypt and her priests.

As for 'unique' it is a modern borrowing from French, and of late '['a]ntique' or '['a]ntic', as Shakespeare has it, has followed in one of its senses the French use. It is a pity in face of Milton's

With mask and ['a]ntique Pageantry,

and it obscures the etymological identity of 'antique' and 'antic', but the old pronunciation is irredeemable. At least the new avoids the homophonic inconvenience.

Greek words of this class used as adjectives mostly follow the same rule, as 'sporadic', 'dynamic', 'pneumatic', 'esoteric', 'philanthropic', 'emetic', 'panegyric'. As nouns the earlier introductions threw the stress back, as 'heretic', 'arithmetic', but later words follow the adjectives, as 'emetic', 'enclitic', 'panegyric'. As for 'politic', which is stressed as we stress both by Shakespeare and by Milton, it must be under French influence, though Skeat seems to think that it came straight from Latin.

STEMS IN -OS. These words agree in being disyllabic, but otherwise they are a tiresome and quarrelsome people. For their diversity in spelling some can make a defence, since 'horror', 'pallor', 'stupor' came straight from Latin, but 'tenor', coming through French, should have joined hands with 'colour', 'honour', 'odour'. The short vowel is inevitable in 'horror' and 'pallor', the long in 'ardour', 'stupor', 'tumour'. The rest are at war, 'clamour', 'colour', 'honour', 'dolour', 'rigour', 'squalor', 'tenor', 'vigour' in the short legion, 'favour', 'labour', 'odour', 'vapour' in the long. Their camp-followers ending in -ous are under their discipline, so that, while 'cl[)a]morous', 'r[)i]gorous', 'v[)i]gorous' agree with the general rule, '[=o]dorous' makes an exception to it. All the derivatives of _favor_ are exceptions to the general rule, for 'favourite' and 'favorable' keep its long _a_. Of course 'l[)a]b[=o]rious' is quite in order, and so is 'v[)a]pid'.

STEMS IN -TOR AND -SOR. These words, when they came through French, threw the stress back and shortened the penultimate, _[=o]r[=a]torem_ becoming _orateur_, and then '[)o]r[)a]tor', with the stress on the antepenultimate. Others of the same type are 'auditor', 'competitor', 'senator', and Shelley has

The sister-pest, congr['e]gator of slaves,

while 'amateur' is borrowed whole from French and stresses its ultima. Trisyllables of course shorten the first vowel, as 'cr[)e]ditor', 'j[)a]nitor'. Polysyllables follow the stress of the verbs; thus '['a]gitate' gives '['a]gitator' and 'comp['o]se' gives 'comp['o]sitor'. To the first class belongs 'circulator', 'educator', 'imitator', 'moderator', 'negotiator', 'prevaricator', with which 'gladiator' associates itself; to the second belongs 'competitor'. Words which came straight from Latin keep the stress of the Latin nominative, as 'creator', 'spectator', 'testator', 'coadjutor', 'assessor', to which in Walton's honour must be added 'Piscator' and 'Venator'. On 'curator' he who decides does so at his peril. On one occasion Eldon from the Bench corrected Erskine for saying 'c['u]r[)a]tor'. 'Cur[=a]tor, Mr. Erskine, cur[=a]tor.' 'I am glad', was the reply, 'to be set right by so eminent a sen[=a]tor and so eloquent an or[=a]tor as your Lordship.' Neither eminent lawyer knew much about it, but each was so far right that he stuck to the custom of his country. On other grounds Erskine might be thought to have committed himself to 't['e]st[)a]tor', if not quite to the 'testy tricks' of Sally in Mrs. Gaskell's 'Ruth'.

STEMS IN -ERO AND -URO. Adjectives of this type keep the Latin stress, which thus falls on the ultima, and shorten or obscure the penultimate vowel, as 'mature', 'obscure', 'severe', 'sincere', but of course '[=a]ustere'. Of like form though of other origin is 'secure'. Nouns take an early stress, as '['a]perture', 's['e]pulture', 'l['i]terature', 't['e]mperature', unless two mutes obstruct, as in 'conj['e]cture'. Of the disyllables 'nature' keeps a long penultima, while 'figure' has it short, not because of the Latin quantity, but because of the French.

The lonely word 'mediocre' lengthens its first vowel by the 'alias' rule and also stresses it. Whether the penultima has more than a secondary stress is a matter of dispute.

STEMS IN -ARI. These words have the stress on the antepenultima, which they shorten, as in 'secular' or keep short as in 'jocular', 'familiar', but of course 'pec[=u]liar'.

_ON CERTAIN GREEK WORDS._

It will have been seen that Greek words are usually treated as Latin. Thus 'crisis' lengthens the penultima under the 'apex' rule, while 'critical' has it short under the general rule of polysyllables. Other examples of lengthening are 'bathos', 'pathos', while the long quantity is of course kept in 'colon' and 'crasis'. For the 'alias' rule we may quote '[=a]theist', 'cryptog[=a]mia', 'h[=o]meopathy', 'heterog[=e]neous', 'pandem[=o]nium', while the normal shortenings are found in 'an[)o]nymous', 'eph[)e]meral', 'pand[)e]monium', '[)e]r[)e]mite'. Ignorance of English usage has made some editors flounder on a line of Pope's:

Yes, or we must renounce the Stagirite.

The birthplace of Aristotle was of course Stag[=i]ra or, as it is now fashionable to transcribe it, Stageira, as Pope doubtless knew, but the editors who accuse him of a false quantity in Greek are on the contrary themselves guilty of one in English. The penultima in English is short whether it was long or, as in 'dynamite' and 'malachite', short in Greek.

There is, however, one distinct class of Greek words in which the Latin rule is not followed. In the sixteenth and seventeenth centuries there were scholars who rightly or wrongly treated the Greek accent as a mark of stress. It is clear that this habit led to an inability to maintain a long quantity in an unstressed syllable. Shakespeare must have learnt his little Greek from a scholar who had this habit, for he writes 'Andr['o]n[)i]cus' and also

I am mis['a]nthr[)o]pos and hate mankind.

Of course all scholars shortened the first vowel of the word, and doubtless Shakespeare shortened also the third. Busby also thus spoke Greek with the result that Dryden in later life sometimes wrote epsilon instead of eta and also spoke of 'Cleom['e]nes' and 'Iphig[=e]n[)i]a'. As a boy at Westminster he wrote

Learn'd, Vertuous, Pious, Great, and have by this
An universal Metempsuchosis.

Macaulay with an ignorance very unusual in him rebuked his nephew for saying 'metam['o]rph[)o]sis', and Dr. Johnson, had he been living, would have rebuked Macaulay. For the sake of our poets we ought to save 'apoth['e][)o]sis', which is in some danger. Garth may perhaps be forgotten,

Allots the prince of his celestial line
An Apotheosis and rights divine,