How can these principal Sounds, after proper Classification, be expressed by us in writing and printing, so as to preserve their physiological Value, without creating new typographical Difficulties?

The results at which we have arrived in the first part of our inquiry are those on which, with very slight and unimportant exceptions, all may be said to agree, who, whether in India or Europe, have attempted to analyse scientifically the elements of human speech. There are, no doubt, some refinements, and some more accurate subdivisions, as will be seen in the extracts given from the Pratisakhyas, which it will be necessary to attend to in exceptional cases, and particularly in philological researches. But, as far as the general physiological outlines of our phonetic system are concerned, we hardly expect any serious difference of opinion.

Widely different opinions, however, start up as soon as we approach the second question, how these sounds are to be expressed in writing. Omitting the different propositions to adopt an Oriental alphabet, such as Sanskrit or Arabic, or the Greek alphabet, or newly invented letters, whether short-hand or otherwise, we shall take it for granted that the Latin alphabet, which, though of Semitic origin, has so long been the armour of thought in the struggles and conquests of civilisation, has really the greatest and most natural claims on our consideration.

There are two principles regulating the application of the Latin page 27 alphabet to our physiological sounds on which there has ben a general agreement since the days of Halhed and Wilkins:

1.	That each sound shall have but one representative letter, and that therefore each letter shall always express the same sound.
2.	That each simple sound shall be expressed by a single lette, and compound sounds by compound letters.

If with these two principles we try to write the forty-fourconsonants of our physiological alphabet by means of the twenty-for consonants of the Latin, it follows that we must add to the latter diacritical signs, in order to make them answer our purpose.

In a theoretical system we might even go a step beyond this, and lay it down as a principle that the same diacritical mark should always express one and the same modification. The advantages which would result from the adoption of such a principle are palpable; but the variety of diacritical marks which it would entail upon us, and the number of new types which would have to be cast to carry it out consistently, must strongly militate against it, particularly in the construction of a Missionary alphabet. Here, as in all branches of Missionary labour, it must be our aim to obtain the greatest results by the smallest means.

Guttural, Palatal, and Dental Tenuis.

The guttural, dental, and labial tenues are naturally expressed by k, t, p.

Guttural, Palatal, and Dental Media.

The modification which changes these tenues into medi æ should consistently be expressed by a uniform diacritical sign attached to k, t, p. For more than one reason, however, we prefer the Latin letters, g, d, b.

It is understood that g, after once being chosen as the representative of the guttural media, like g in gun, whatever vowel may follow, can never be used promiscuously both for the guttural and the palatal media, as the English g in gun and gin.

How to express Aspirates?

These types have been cut many times since Count Volney founded his prize at the French Academy for transcribing Oriental alphabets, and even before his time. They exist at Berlin, Paris, Leipzig, Darmstadt, Petersburg, and several other places. They have been cut in different sizes and on different bodies. Still the difficulty of having them at hand when required, making them range properly, and keeping always a sufficient stock, has been so great even in places like London, Paris, and Berlin, that their adoption would defeat the very object of our alphabet, which is to be used in Greenland as well as in Borneo, and is to be handled by unexperienced printers even in the most distant stations, where nothing but an ordinary English font can be expected to exist. In our Missionary alphabet we must therefore have no dots, no hooks, no accents, no Greek letters, no new types, no diacritical appendages whatsoever. No doubt, Missionary Societies might have all these letters cut and cast on as many sizes and bodies as necessary. Punches or fonts might be sent to the principal Missionary stations. But how long would this last? If a few psalms or catechisms had to be printed at Bangkok, and if there were no hooked letters to represent the aspirated palatal sound by a single type (k″), is it likely that they would send to Calcutta or London for this type, which, after it arrived, might perhaps be found not to range with the rest? It is much more likely that, in the absence of the type prescribed by the Missionary Societies at home, each missionary would find himself thrown on his own resources, and different alphabets would again spring page 29 up in different places. Besides, our alphabet is not only to be an alphabet of missionaries. In time it is to become the alphabet of those tribes and nations whose first acquaintance with writing will be through the Bible translated into their language and transcribe in a rational alphabet. Fifty or a hundred years hence, it may be the alphabet of all the civilised nations of Africa, Australia, and the greater part of Asia. Must all the printers of Australian advertisements, the editors of African newspapers, the publishers of Malay novels or Papua primers, write to Mr. Watts, Crown Court, Temple Bar, for new sorts of dotted and hooked letters? I do not sty it is impossible; but many things are possible, and still not practical; and these new hooked and dotted types seem to me decidedly to belong to this class.

In questions of this kind, no harm is done if principles are sacrificed to expediency; and I therefore propose to write the aspirate letters, as all English and most French and German scholars have written them hitherto, by

kh, th, ph, gh, dh, bh.

What do we lose by this? The spiritus asper (') is after all but a faintly disguised II, changed into I- and -I, for asper and lenis, and then abbreviated into 'and'. Besides, the languages where these simple aspirates occur are not many; and in India, where they are of most frequent use, the phonetic system is so carefully arranged that there no ambiguity can arise whether kh be meant for an aspirated guttural tenuis or for k followed by the semi-vowel h. If the semi-vowel h comes in immediate contact with k, k + h is changed into g + gh, or a stop (virama) has to be put after the k. This might be done where, as in discussing grammatical niceties, it is desirable to distinguish between kh and k-h. The missionary, except in India, will hardly ever suffer from this ambiguity; and if the scholar should insist on its being removed, we shall see immediately how even the most delicate scruples on this point could be satisfied.

There is still, if we examine the alphabets hitherto proposed or adopted, a whole array of dots and hooks, which must be eliminated, or at least be reduced, as far as possible; and though we might, after gaining our point with regard to the h, get through gutturals, dentals, and labials, we still have new and more formidable enemies to encounter in the palatals and linguals.

How to express Palatals?

Palatals are modifications of gutturals, and therefore the most natural course would be to express them by the guttural series, adding only a line or an accent or a dot, or any other uniform diacritical sign to indicate their modified value. So great, however, has been the disinclination to use diacritical signs, that in common usage, where the palatal tenuis had to be expressed, the most anomalous expedients have been resorted to in order to avoid hooks or dots. In English, to represent the Sanskrit palatal tenuis, ch has been used; and as the h seemed to be too much in the teeth of all analogy, the simple c even has been adopted, leaving ch for the aspirated palatal. On the same ground, the Germans write tsch for the palatal tenuis, and tschh for the aspirate. The French write tch and tchh. The Italians do not hesitate to use ci for the tenuis, though I do not see how they could express the corresponding aspirate. The Russians recommend their Ч; and the Brahmans would probably recommend a Sanskrit type. Still all, even the German tschh, are meant to represent simple consonants, which, as in Sanskrit, would not make a preceding short vowel long. That in English the ch, in Italian ci, and in German tsch, have a sound very like the palatal tenuis, is of course a mere accident. In English the ch is not always sounded alike; and its pronunciation in the different dialects of Europe varies more than that of most letters. Besides, our alphabetic representative of the palatal sound is to be pronounced and comprehended, not by a few people in Germany or Italy, but by all the nations of Africa and Australia. Now to them the ch would prove deceptive; first, because we never use the simple c (by this we make up for the primary alphabetical divorce introduced by the libertus of Spurius Carvilius Ruga), and, secondly, because the h would seem to indicate the modification of the aspirate.

The natural way of writing the palatals, so as not to obscure their close relationship to the gutturals, would be, k, kh, g, gh.

But here the same difficulty arises as before. If the dots or marks are printed separately, the lines where these dots occur become more distant than the rest. For one such dotted letter the compositor has to compose a whole line of blanks. These will shift, particularly when there are corrections, and the misprints are endless. page 31 In Tumour's edition of the Mahavansa, which is printed with dotted letters, we get thirty-five pages quarto of errata to about a hundred pages of text. But they might be cast on one body. True, they might be—perhaps they will be. At all events they have been; and Volney offered such types to anybody that would ask for them. Still, when I inquire at a press like the University press of Oxford, they are not forthcoming. We must not expect that what is impossible in the nineteenth century at Oxford, will be possible in the twentieth century at Timbuktu.

Now the difficulty, so far as I can see, was solved by a compositor to whom I sent some MS., where each palatal letter was marked by a line under it. The compositor, not knowing what these lines meant, took them for the usual marks of italics, and I was surprised to see that this answered the purpose, saved much trouble and much expense, and, on the whole, did not look badly. As every English font includes italic letters, the usefulness of these modified types for our Missionary alphabet "springs to the eyes," as we say in German. They are sufficiently startling to remind the reader of their modified pronunciation, and at the same time they indicate, as in most cases they ought, their original guttural character to the reflecting philologist. As in ordinary books italics are used to attract attention, so also in our alphabet. Even to those who have never heard the names of guttural and palatal letters, they will show that the k is not the usual k. Persons in the slightest degree acquainted with phonetics will be made aware that the k is, in shape and sound, a modification of the k. All who admit that palatals are modifications of gutturals would see at once that the modification intended by k could only be the palatal. And as to the proper pronunciation of the k, as palatal tenuis, in different dialects, people who read their own language expressed in this alphabet will never hesitate over its pronunciation. Others must learn it, as they now learn the pronunciation of Italian ci and chi, or rest satisfied to know that k, stands for the palatal tenuis, and for nothing else. Sooner or later this expedient is certain to be adopted. Thus we get, as the representatives of the palatals,

k, kh, g, gh.

Now, also, it will appear how we can avoid the ambiguity before page 32 alluded to,—whether the h of aspirated consonants expresses their aspirated nature or an independent guttural semi-vowel or flatus. Let the h, where it is not meant as a letter, but as a diacritical sign, be printed as an italic h, and the last ground for complaint will vanish. Still this is only needful for philological objects; for practical purposes the common h may remain.

In writing, the dots or lines under the palatals will have to be retained. Still they take too much time thus employed to allow us to suppose that the Africans will retain them when they come to write for themselves. They will find some more current marks; as, for instance, by drawing the last stroke of the letter below the line. In writing, however, anybody may please himself, so long as the printer knows what is intended when he has to bring it before the public. As a hint to German missionaries, I beg to say that, for writing quickly in this new alphabet, they will find it useful in manuscript notes to employ German letters instead of italics.

An accidental, though by no means undesirable, advantage is gained by using italics to express the palatals. If we read that Sanskrit vâch (or vâtch, or vâtsch) is the same as Latin vox, but that sometimes vâch in Sanskrit is vâk or vac, the eye imagines that it has three different words to deal with. By means of italics, vâk and vâk are almost identical to the sight, as kirk and kurk (church), would be if English were ever to be transcribed into the missionary alphabet. The same applies to the verb, where the phonetic distinction between vakmi, vakshi, vakti, can thus be expressed without in any way disguising the etymological identity of the root. It would be wrong if we allowed the physiological principles of our alphabet to be modified for the sake of comparative philology; but where the phonetic changes of physiological sounds and the historical changes of words happen to run parallel, an alphabet, if well arranged, should be capable of giving this fact clear expression.

If the pronunciation of the palatals is deteriorated, they sometimes take the sound of tch, ts, s, sh, or even th. Ccelum () becomes Italian cielo; where the initial sound is the same as in church (kirk). In old Friesic we have "tzaka" instead of English "check." In French, "ciel" is pronounced with an initial sharp dental s; " chose," with an initial sharp palatal s. In Spanish, the pronuncia- page 33 tion of a c before e and i is that of the English th. In these cases when we have to deal with unwritten languages, the sounds, whether simple or double, should be traced to their proper phonetic category, and be written accordingly. It will be well, however, to bear in mind that pronunciation may change with time and vary in different places, and that the most general representation of these sounds by palatals or italicized gutturals will generally prove the best in the long run.

It must be clear that, with the principles followed hitherto, it would be impossible to make an exception in favour of the English j as representative of the palatal media. It would be a schism in the whole system, and would besides deprive us of those advantages which comparative philology derives from a consistent representation of modified sounds: that Sanskrit yuga () is derived from " yug," to join, would be intelligible to everybody; while neither the German, to whom j is y, nor the Frenchman, nor the Spaniard would see the connexion between j and g.

The wish to retain the j is natural with Missionary Societies. It would enable us to spell uniformly the name of our Lord—and in all the translations of the Bible which the pious zeal of the mother country is now sowing over the virgin soil of Africa, Australia, and Asia, that one name at least would stand unaltered and uncorrupted in all tongues and all ages. But we may consider this from another point of view. As with other words, and with many of the most sacred in our own language, their full and real meaning seems to grow more clear and distinct the more the material body of the words changes and decays, and the more their etymological meaning becomes dim and forgotten, so will it he with the name of our Lord. Let the name grow and change and vary in all the tongues of the earth, and the very variety of the name will proclaim the unity of Him who has promised to all tongues the gift of His Holy Spirit. And would it avail, even if now we insisted on this point? A thousand years ago, and all the nations of Europe wrote and pronounced this name uniformly; but at the present day there are hardly two languages where the name is pronounced exactly alike; and in several the spelling has followed the pronunciation. It will ultimately be the same in Africa, whatever we do at present. But if page 34 an exception is here to be made, let it be a single exception, while we retain the regular notation for every other word in which the pure palatal media occurs.

How to express Linguals?

The linguals, as modifications of the dentals, have been hitherto written by common consent as dentals with dots or lines. In writing, this method must be retained, though no doubt a more current form will soon grow up if the alphabet is used by natives. They will probably draw the last stroke of the t and d below the line, and connect the body of the letter with the perpendicular line below. The linguals, therefore, will be, t, th, d, dh; only here also the printer will step in and convert the dotted or underlined letters into italics, t, th, d, dh.

I am at a loss how to mark that peculiar pronunciation of the dental aspirate, whether tenuis or media, which we write in English simply by th. It is not of frequent occurrence; still it occurs not only in European, but in Oriental languages,—for instance, in Burmese. If it occurs in a language where no trace of the pure dental aspirate remains, we might safely write th (and dh) or th (and dh), as we do in English. The Anglo-Saxon letters and ð would be very convenient; but how few fonts, even in England, possess these forms! Again, h and zh, and even and , have been proposed; but they are liable to still stronger objections. Where it is necessary to distinguish the aspirated th and dh from the assibilated, I propose for the latter a dot under the h (th and dh). But I think th and dh will, on the whole, be found to answer all practical purposes, if we only look to people who have to write and read their own language. Philologists, whatever we attempt, cannot be informed of every nicety and shade in pronunciation by the eye. They must learn from grammars or from personal intercourse in what manner each tribe pronounces its dental aspirate; and comparative philology will find all its ends answered if th represents the organic dental aspirate, until its pronunciation deteriorates so far as to make it a flatus or a double consonant. In this case the Missionary also will have to write it Is, or ss, or whatever sound he may happen to hear.

How to express the Nasals?

In each of these five classes we have now to look for an exponent of the nasal.

Where the nasal is modified by the following consonant, it requires no modified sign, for reasons explained in the first part of our essay. The nasal in sink and sing is guttural; in inch and injure, palatal; in hint and bind, dental; in imp and dumb, labial.

But where these nasals occur at the beginning of words or at the end of syllables, each must have its own mark. Let the dental nasal be n, the labial nasal m, the lingual nasal n. Where the guttural nasal is really so evanescent as not to bear expression by ng, we must write n and a dot after it (n.), which makes no difficulty in printing, and will very rarely occur. What we call the palatal n is generally not a simple but a compound nasal, and should be written ny. For transliterating, however, we want a distinct sign, because the palatal nasal exists as a simple type in Sanskrit, and every single type must be transliterated by a single letter. Here I should propose the Spanish ñ.

The lingual n occurs in Sanskrit only. Its character is generally determined by lingual letters either following or preceding. Still, where it must be marked in Sanskrit transliterations, let it be represented by an italic n.

How to express the Semi-vowels?

The Latin letters which naturally offer themselves as the counterparts of the semi-vowels, are 'h, y, r, l, and w.

The delicate sound of the guttural semi-vowel occurs very rarely in Arian languages. In Semitic dialects, however, the has usually been considered as the guttural semi-vowel. In Hebrew it is sometimes not pronounced at all, or, as we should say, it is changed into the flatus lenis; whence, in the Arabic alphabet, to remove this ambiguity and to show in every word the full or weak pronunciation of the guttural semi-vowel, the was split in two: the one, the little more than the flatus lenis; the other, the , the hollow guttural semi-vowel which only a Semitic throat is able to utter, and which comes very near to the guttural flatus asper as heard in "loch."

The palatal semi-vowel is usually transcribed in Germany by j, which, as far as archæological arguments go, would certainly be the most appropriate sign to represent the semi-vowel corresponding to the palatal vowel i. As, however, the j is one of the most variously pronounced letters in Europe, and as in England it has been usual tc employ it as a palatal media, it is better to discard it altogether from our alphabet, and to write y.

The lingual semi-vowel is r; if in some dialects the r is pronounced very near to the throat, this might be marked by an italic r, or rh.

The dental semi-vowel is written 1. The mouillé sound of 1 may be expressed by an italic l.

Where the labial semi-vowel is formed by the lips, let it be written w. More usually it is formed by the upper lip and the edge of the lower teeth. It then becomes what the Hindus call a labio-dental semi-vowel, but is hardly to be distinguished from the labial flatus lenis.

How to express the Flatus (Sibilants)?

As the unmodified flatus, or, as it should more properly be called, the spiritus asper and lenis, can only occur before a vowel, the printer will find no difficulty in representing these two sounds by the usual signs 'and ' placed before or over the vowel which follows. At the beginning of words there could be no reasonable objection to this mode of representing the very slight and hardly consonantal sound of the spiritus asper and lenis. But it will take some time before our eyes are accustomed to it in the middle of words. In such cases the Greeks did not mark it. They wrote , chariot, page 37 but , with beautiful chariots; they wrote , man, but , manliness. Nor in fact does there seem to be any necessity for marking the spiritus lenis in the middle of words. Every vowel beginning a syllable has necessarily the spiritus lenis; as going, seeing. As to the spiritus asper, which we have in "vehement," "vehicle," I fear that "ve'ement," "ve'icle," will be objected to by the printer. If so, we have still the h as a last resource to express the spiritus asper in this position.

The guttural flatus asper, as heard in loch, might be expressed by an Italic h. The flatus lenis cannot be distinguished in pronunciation from the guttural semi-vowel, and has therefore never received an alphabetical exponent. If it.should be necessary, however, to assign a type to this physiological category, we should be obliged to write the flatus asper by 'h, and the flatus lenis by 'h.

The dental flatus sibilans, pronounced sharp as in "sin" or "grass," has, of course, the best claims on the letter s as its representative. Its corresponding soft sound, as heard in please or zeal, is best expressed by z; only we must take care not to pronounce it like the German z. The more consistent way of expressing the sonant flatus would be to put a spiritus lenis over the s. This, however, would hardly be tolerated, and would be against the Third Resolution of our alphabetical conferences, where it was agreed that only after the Roman types, and the modifications of Roman types as supplied by common fonts (capitals, italics, &c.), had been exhausted, diacritical signs should be admitted into the standard alphabet.

As all palatals are represented by italics, the palatal sibilant would naturally be written with an italic s. This would represent the sharp sound as heard in "sharp " or "chose." The soft palatal sibilant would have the same exponent as the soft dental sibilant, only changed into italics (z). This would be the proper sign for the French sound in "je," "genou," and for the African soft palatal sibilant, which, as Dr. Krapf, Mr. Tutschek, and Mr. Boyce remark, will never be properly pronounced by an adult European.

Where it is necessary to express the original, not yet assibilated, palatal flatus, which is heard in könig and kön'ge, an italic y, with the spiritus asper arrd lenis, would answer the purpose ('y and y)

The labial flatus should be written by f. This is the shap flatus, page 38 as heard in "life" and "find." The soft labial flatus ought consistently to be written as f with a spiritus lenis. But here again I fear we must sacrifice consistency to expediency, and adopt that sign with which we are familiar, the Latin v. As we express the labial semivowel by w, the v is still at our disposal, and will probably be preferred by the unanimous votes of missionaries and printers.

The lingual flatus is a sound peculiar to Sanskrit, and, owing to its hollow guttural pronunciation, it may be expressed there, as it has been hitherto, by s followed by the guttural h (sh). The Sanskrit knows no soft sibilants; hence we require but one representation for the lingual sh.

Spiritus asper:'.

Spiritus lenis: '.

Although these exponents of the physiological categories of articulated sound have not been chosen because their present pronunciation in English, or French, or German is nearest to that physiological category which each has to represent, still, as we have avoided letters of which the pronunciation fluctuates very much (such as c, j, x, q), it will be found, on the whole, that little violence is done by this alphabet to the genius of any of these languages, and that neither an Englishman, nor a German, nor a Frenchman will ever feel much hesitation as to how any one of our letters should be pronounced.

Vowels.

The pronunciation of the vowels is more liable to change than that of the consonants. Hence we find that literary languages, which page 39 retain their orthography in spite of changes in pronunciation have no scruple in expressing different sounds by the same sign; or, where two originally different vowels have sunk down to one and the same intermediate sound, we see this same sound expressed often by two different vowels. In the selection, therefore, of letters to express the general vowel sounds of our physiological alphabet, we can pay less attention to the present value of each vowel sign in the spoken languages of Europe than we did even with the consonants. And as there it was impossible, without creating an unwieldy mass of consonantal signs, to express all the slight shades of pronunciation by distinct letters, we shall have to make still greater allowance for dialectical varieties in the representation of vowels, where it would be hopeless should we attempt to depict in writing every minute degree in the sliding scale of native or foreign pronunciation.

The reason why, in most systems of phonetic transcription, the Italian pronunciation of vowels has been taken as normal, is, no doubt, that in Italian most vowel signs have but one sound, and the same sound is generally expressed by one and the same vowel. We propose, therefore, as in Italian, to represent the pure guttural vowel by a, the pure palatal vowel by i, and the pure labial vowel by u.

Besides the short a, we want one, or according to others, two graphic signs to represent the unmodified sound of the vocal breathing, which may be deflected from its purity by a slight and almost imperceptible palatal or labial pressure. These are the sounds which we have in "birch" and "work," and which, where they must be distinguished, we propose to write ĕ and ŏ. As we do not want the signs of and to mark the quantity of vowels, we may here be allowed to use this sign to indicate indistinctness rather than brevity.

In most languages, however, one sign will be sufficient to express this primitive vowel; and in this case the figure 0 has been recommended as a fit representative of this undetermined vowel.

Among the languages which have an alphabet of their own, some, as, for instance, Sanskrit, do not express these sounds by any peculiar sign, but use the short a instead. Other languages express both sounds by one sign; for instance, the Hebrew Shewa, the pronunciation of which would naturally be influenced, or, so to say, coloured either by the preceding or the following letter. Other idioms again, page 40 like Latin, seem to express this indistinct sound by e, i, o, or u. Besides the long e in res and the short e in celer, we have the indistinct ĕ in words like adversum and advorsum, septimus and septumus, where the Hindus write uniformly saptama, but pronounced it probably with vowels varying as in Greek and Latin. Besides the long o in odi, and the short o as in moneo, we have the indistinct o or u in orbs or urbs, in bonom or bonum. In Wallachian, every vowel that has been reduced to this obscure, indefinite sound, is marked by an accent, a, e, i, b, u; but if Wallachian is written with Cyrillic letters, the ' Yerr' ('b) is used as the uniform representative of all these vowels. In living languages one sign, the figure o, will be found sufficient, and in some cases it may be dispensed with altogether, as a slight Shewa sound is necessarily pronounced, whether written or not, in words such as mil-k, mar-sh, el-m, &c. The marks of quantity,? and are superfluous in our alphabet; not that it is not always desirable to mark the quantity of vowels, but because here again, as with the dotted consonant, a long syllable can be marked by the vowel in italics, while every other vowel is to be taken as short. Thus we should write in English bath, bar, but ass, bank; ravine, and pin; but (i.e. boot), and butcher. We should know at once that a in bath is long, while in ass it is short.

All compound vowel sounds should be written according to the process of their formation. Two only, which are of most frequent occurrence, the guttural short a, absorbed by either i or u, might perhaps be allowed to retain their usual signs, and be written e and o, instead of ai and au. The only reason, however, which can be given for writing e and o, instead of ai and au, is that we save a letter in writing; and this, considering how many millions of people may in the course of time have to use this alphabet, may be a saving of millions and millions of precious seconds. The more consistent way would be to express the gutturo-palatal sound of the Italian e by ai, the a being short. The French do the same in "aimer," while in English this sound is expressed by ey in prey, by ay in pray, by a in gate, and by ai in sailor. The gutturo-labial sound of the Italian o should be written au, which the French pronounce o. For etymological purposes also this plan would be preferable, as it frequently happens that an o (au), if followed by a vowel, has to be pronounced page 41 av. Thus in Sanskrit bhu, to be, becomes bhau (pronounced bho), and if followed by ami, it becomes bhav-ami, I am.

The diphthongs, where the full or long guttural a is followed by i and u, should be written ai and au. "To buy" would have to be written bai; to bow, bau. Whether au coalesce entirely, as in German, or less so, as in Italian, is a point which in each language must be learned by ear, not by eye.

Most people would not be able to distinguish between ai and ei. Still some maintain that there is a difference; as, for instance, in German kaiser and eis. Even in English the sound of ie in "he lies" is said to be different from that of "he lies." Where it is necessary to mark this distinction, our diagram readily supplies ai and ei.

The diphthong eu is generally pronounced so that the two vowels are heard in succession, as in Italian Europa. Pronounccd more quickly, as, for instance, in German, it approaches to the English sound of oy in boy. According to our diagram, we should have to write ĕi and ĕu; but ei and eu will be preferable for practical purposes.

The same applies to the diphthong ŏi. Here, also, both vowels can still be heard more or less distinctly. This more or less cannot be expressed in writing, but must be learned by practice.

The last diphthong, on the contrary, is generally pronounced like one sound, and the deep guttural O seems to be followed, no by the vowel u, but only by an attempt to pronounce this vowe, which attempt ends, as it were, with the semi-vowel w, instead of the vowel. In English we have this sound in bought, aught, saw; and also in fall and all.

The proper representation of these diphthongs would be ŏi and ŏu; but oi and ou will be found to answer the purpose as well except in philological works.

For representing the broken sounds of a, o, u, which we lave in German väter, höhe, güte, in the French prêtre, pcu, and une, but which the English avoids as sounds requiring too great an efort, no better signs offer themselves than ä, ö, ü. They are objecionable because they are not found in every English font. For the I'ataric languages a fourth sound is required, a broken or soft i. Tis, too; we must write ï.

The Sanskrit vowels, commonly called lingual and dental, are best expressed by ri and li, where, by writing the r and I as italics, no ambiguity can arise between the vowels ri and li, and the semi-vowels r and I, followed by i. Instead of i, ĕ also or the figure O may be used.

Thus have all the principal consonantal and vowel sounds been classified physiologically and represented graphically. All the distinctions which it can ever be important to express have been expressed by means of the Roman alphabet without the introduction of foreign letters, and without using dots, hooks, lines, accents, or any other diacritical signs. I do not deny that for more minute points, particularly in philological treatises, new sounds and new signs will be required. In Sanskrit we have Visarga and the real Anusvara (the Nasikya), which will require distinct signs (h, m) in transliteration. In some African languages, clicks, unless they can be abolished in speaking, will have to be represented in writing. On points like these an agreement will be difficult, nor would it be possible to provide for all emergencies. It is an advantage, however, that we still have the c, j, and x at our disposal to express the dental, palatal, and lateral clicks. Further particulars on this and similar points I must reserve for a future occasion, and refer the reader, in the mean time, to the very able article of the Rev. L. Grout, alluded to before. But I cannot leave this subject without expressing at least a strong hope that, by the influence of the Missionaries, these brutal sounds will be in time abolished, at least among the Kaffirs, though it may be impossible to eradicate them in the degraded Hottentot dialects. It is clear that they are not essential in the Kaffir languages, for they never occur in Sechuana and other branches of the great Kaffir family.

If uniformity can be obtained with regard to the forty-four consonantal and the twenty-four vocal sounds, which are the principal modulations of the human voice fixed and sanctioned in the history of language, so far as it is known at present; if these sounds are always accepted, as defined above, solely on physiological grounds, and henceforth expressed in those letters alone which have been allotted to them solely for practical reasons, a great step will have been made towards facilitating the intellectual intercourse of mankind and spreading the truths of Christianity.

But the realisation of this plan will mainly depend, not on ingenious arguments, but on good-will and ready co-operation.