significant (p. 18). Thus they involve the division of the palate into three parts only, the minuter intermediate positions involved in Jespersen's symbols being indicated by the modifiers, which are graphically subordinated to the elementary symbols. So in this respect the Organic alphabet partially adopts the digraph or analphabetic principle.

Again, in a practical alphabet, the distinctions of nature must often be exaggerated so that there may be no hesitation in distinguishing the symbols of similar sounds. From this point of view Jespersen's objection to Bell's symbolizing consonants and vowels on different principles, so that, for instance, there is no resemblance between the symbols of lowered (j) and non-syllabic (i), appears of little weight. The real objection here seems to be that Bell confuses analysis with synthesis. But, again, if it is more practical and convenient to embody such distinctions as vowel and consonant, syllabic and nonsyllabic, in the elementary analytic symbols, then he is justified in doing so till some one else hits on an arrangement which is more scientific and as practical.

Universal Alphabet not suited for Connected

Writing

A universal notation is, in the nature of things, generally used only to write a few words at a time, sometimes only a single sound. In writing connected texts in one particular language, an alphabet of the Broad Romic type is infinitely more convenient: all the learner has to do is to associate each Broad Romic symbol with the pronunciation of the corresponding Narrow Romic, Organic, or Analphabetic symbol of the sound in question, so that, for instance, when he meets (i) in his texts, he knows that it stands for the high front wide-or whatever shade of sound it is-in the language he is studying.

But it is evidently a great help to the learner-especially if he has not a teacher-to have his texts accompanied by a minutely accurate notation for at least the first page or two. Here an analphabetic notation is perfectly useless.

The advantages of the Roman alphabet for connected transcription are evident: it is an alphabet which has been developed partly by a slow process of spontaneous evolution, partly by conscious reforms and endless experiments.

But it has many defects. From a mechanical point of view, its worst defect is want of compactness (p. 15). În a universal scientific alphabet like the Organic, a certain amount of sprawliness is inevitable; but in a practical alphabet, which has to supply only a limited number of characters, it is an inexcusable defect.

Again, although our script or running-hand alphabet is fairly quick to write, it ought certainly to be quicker than it is. In most of the languages which use the Roman alphabet speed is further impeded by diacritics, such as the accents in French. Even in English the dot over the i and j wastes much time.

Superiority of Phonetic Shorthand

These and other considerations point clearly to the adoption of a system of phonetic shorthand on a general basis capable of being adapted to the special requirements of each language. As the basis of such a shorthand would be necessarily quite independent of the Roman alphabet, the danger of confusion between phonetic and nomic spellings would be reduced to a minimum. The introduction of a phonetic shorthand would, at the same time, be the real solution of the problem of spelling-reform. Lastly, all modern systems of shorthand are based more or less on organic or acoustic associations: they all show some connection between the form of the symbols and the sounds they represent, although, of course, in a practical system of writing theoretical consistency must always yield to considerations of speed and convenience.

Speed. The term 'shorthand' is, in itself, only a relative one. Our ordinary script is a shorthand, if compared with the Roman capitals out of which it developed. The highest development of shorthand as regards speed of writing is, of course, reporting shorthand, whose definite aim is to enable the writer to keep up with a moderately fluent speaker: that is, it must be capable of being written at the rate of about 150 words a minute, which is five times as much as the rate of quick longhand writing.

As speech would outrun the quickest fingers, if every syllable -not to speak of every sound-had to be indicated, if only by a single stroke, high speed necessarily involves contraction -the wholesale omission of vowels, syllables, or even words

the result being generally unintelligible to the writer himself unless copied out into longhand immediately after being written.

Distinctiveness.-A system of shorthand which is to take the place of longhand and retain the latter's advantages must, on the other hand, subordinate speed to legibility. For linguistic purposes it must be more than legible: it must be phonetically distinctive, that is, it must be capable of being transcribed accurately into such a notation as Broad Romic. In its contractions, too, it must be rigorously distinctive: each word, however much contracted, must have its own outline, by which it can be recognized immediately and with certainty without any guessing by the context.

All we can expect, then, from this point of view, is a system of writing as much shorter and more compact than ordinary longhand as the requirements of distinctness and legibility will allow. None of the three systems most in use at the present time-Pitman's in England and America, and the German systems of Gabelsberger and Stolze on the Continent-can be said fully to meet these requirements: they all sacrifice efficiency to brevity, the brevity being often only apparent.

My Current Shorthand is an attempt to supply this want (A Manual of Current Shorthand, Oxford, 1892). In the preface to the Manual I sum up the characteristic features of the system as follows:

1. It is the first workable pure script [as opposed to geometric] shorthand that has been brought out in England.

2. It affords the first satisfactory solution of the vowel problem, by providing separate symbols for them, which, though joined to the consonants, are subordinated to them, so that the vowels can be omitted without altering the general appearance of the word.

3. It is the first system which makes a systematic use of projection above and below the line of writing to indicate the different classes of consonants.

4. It provides a purely orthographic and a purely phonetic style of writing for concurrent use.

5. It discards not only thick and thin, but all other sham distinctions.

6. It is rigorously linear, so that it can be used for all the purposes of ordinary longhand.

7. It could be printed from moveable types with comparative ease.

8. It is on a strictly syllabic and alphabetic basis.

Modified Nomic Spelling

As already remarked (p. 14), the defects of a comparatively phonetic orthography such as that of Italian or German can be easily remedied without substituting a new orthography.

A nomic orthography can be supplemented in the six ways enumerated on p. 14.

If

Of these methods, the use of diacritics is peculiarly applicable to the orthographies of dead languages, especially those in which it is desirable to reproduce the varying spellings of the original manuscripts, as in printing Old English or Old Irish texts. It is often a great advantage to have such texts printed in such a form as to enable the reader to see at a glance what is the original manuscript spelling, while at the same time he is supplied with the additional information required for the discrimination of the distinctive sounds of the language as far as they have been determined with any degree of probability. Thus in Old English there are two sounds of c, namely (k) and (c), the former being sometimes written k in the manuscripts. If our manuscript has k, we print it so; if the manuscript has c, we print it c when it stands for (k), ċ when it stands for (c). we were constructing a new phonetic transcription of Old English, we should transliterate the two sounds by k and c respectively, as being more distinct and convenient than cand c. But this is inadmissible if we wish faithfully to preserve the evidence of the manuscripts. So also it is better to mark long vowels in Old English with ( than by doubling-which we might prefer in a free phonetic transcription-or the addition of ('), for quantity is occasionally marked in these last ways in the manuscripts, but never by the macron or circumflex, either of which may therefore be employed. Hence such spellings as kene, cyning, ciese, éhese, in my Anglo-Saxon Reader serve both to indicate the exact pronunciation of these words, and to allow the reader to infer that the original manuscript spellings are kene, cyning, ciese, ciese.

Although diacritics have peculiar advantages as regards restitution of the original manuscript spellings, there is no objection to substituting other letters which do not occur in

the ordinary orthography of the language in question. Thus if k never occurred in Old English manuscripts, there would be no harm in using it instead of c, so that the other sound could be represented by simple c. In the same way we could substitute for, or use it to distinguish the open o in lond as opposed to the close o in on, boren, for none of these substitutions would hinder the recovery of the manuscript spelling. Italics are often very convenient for such discriminations of pairs of sounds.

Italics are specially useful in indicating silent letters, such as the final e in many words in Chaucer's English. As silent letters do not occur in Old English, italics can be used there to mark the omission by the manuscript of a letter required by strict phonetic spelling, as in mann for the manuscript spelling

man.

Even modern English might be written phonetically in this way. Thus through, though, thy might be written (through, though, Thy). But any such method breaks down practically with such an orthography as the English; and it is much simpler in the end to start with an entirely new phonetic spelling, as distinct from the nomic spelling as possible.