é and diphthongs, because they combine the conditions of a and i, and of a and u, we might call the Sanskrit ch a consonantal diphthong, though even this would lead to the false supposition that it was necessarily a double letter, which it is not. That the palatal articulation may be simple is clearly seen in those languages where, as in Sanskrit, both ancient and modern, ch leaves a short vowel that precedes it short, whereas a double consonant would raise its quantity.

Few Sanskrit scholars acquainted with the Prâtisâkhyas, works describing the formation of letters, would venture to speak dogmatically on the exact pronunciation of the so-called palatal letters at any definite period in the history of ancient Sanskrit. They may have been pronounced as they are now pronounced, as consonantal diphthongs; they may have differed from the gutturals no more than k in kaw differs from k in key; or they may have been formed by raising the convex part of the tongue so as to flatten it against the palate, the hinder part being in the k, and the front part in the y position. The k, as sometimes heard in English, in kind, card, cube, cow, sounding almost like kyind, cyard, cyube, cyow, may give us an idea of the transition of k into ky, and finally into English ch,- a change analogous to that of t into ch, as in natura, nature, or of d into j, as in soldier, pronounced soljer, diurnale changed to journal. In the northern dialects of Jütland a distinct j is heard after k and g if followed by æ, e, o, ö; for instance, kjæv', kjær, gjekk, kjerk, skjell, instead of kæv', kær, &c.1 However that may

1 See Kuhn's Zeitschrift, xii. 147.

be, we must admit, in Sanskrit and in other languages, a class of palatals, sometimes modifications of gutturals, sometimes of dentals, varying no doubt in pronunciation, not only at different periods in the history of the same language, but also in different localities; yet sufficiently distinct to claim a place for themselves, though a secondary one, between gutturals and dentals, and embracing, as we shall see, the same number of subdivisions as gutturals, dentals, and labials.

It is not always perceived that these three consonants k, t, p, and their modifications, represent in reality two quite different effects. If we say ka, the effect produced on the ear is very different from ak. In the first case the consonantal noise is produced by the sudden opening of the tongue and palate; in the second, by their shutting. This is still clearer in pa and ap. In pa you hear the noise of two doors opening, in ap of two doors shutting. In empire you hear only half a p; the shutting takes place in the m, and the p is nothing but the opening of the lips. In topmost you hear likewise only half a p; you hear the shutting, but the opening belongs to the m. The same in uppermost. It is on this ground that mute letters have sometimes been called dividuæ, or divisible, as opposed to the first class, in which that difference does not exist; for whether I say sa or as, the sound of s is the same.

Soft Checks, or Media.

We should now have finished our survey of the alphabet of nature, if it was not that the consonantal stops k, t, p, are liable to certain modifications, which,

as they are of great influence in the formation of language, deserve to be carefully considered. What is it that changes k into g and ng, t into d and n, p into b and m? B is called a media, a soft letter, a sonant, in opposition to p, which is called a tenuis, a hard letter, or a surd. But what is meant by these terms? A tenuis, we saw, was so called by the Greeks in opposition to the aspirates, the Greek grammarians wishing to express that the aspirates had a rough or shaggy sound,1 whereas the tenues were bald, slight, or thin. This does not help us much. "Soft" and "hard" are terms which no doubt express the outward difference of p and b, but they do not explain the cause of that difference. "Surd" and "sonant" are apt to mislead; for, as both p and b are classed as mutes, it is difficult to see how a mute letter could be sonant. Some persons have been so entirely deceived by the term sonant, that they imagined all the so-called sonant letters to be necessarily pronounced with tonic vibrations of the chordæ vocales.2 This is physically impossible; for if we really tried to intone p or b, we should either destroy the p and b, or be suffocated in our attempt at producing voice. Both p and b, as far as tone is concerned, are aphonous or mute. But b differs from p in so far as, in order to pronounce it, the breath is for a moment checked by the glottis, just as it was in pronouncing v instead of f. What, then, is the difference between German w and b? Simply that in the former no contact takes place,

1 Brücke, p. 90. Aspiration, p. 103.

τῷ πνεύματι πολλῷ, Dion Hal. R. von Raumer, Die

2 Funke, p. 685. Brücke, Grundzüge, pp. 7, 89.

and hence no cessation of breath, no silence; whereas the mute b requires contact, complete contact, and hence causes a pause, however short it may seem, so that we clearly hear the breath all the time it is struggling with the lips that shut in upon it. We may now understand why the terms soft and hard, as applied to b and p, are by no means so inappropriate as has sometimes been supposed. Czermak, by using his probe, as described above, found that hard consonants (mutæ tenues) drove it up much more violently than the soft consonants (mutæ mediæ).1 The normal impetus of the breath is certainly checked, subdued, softened, when we pronounce b; it does not strike straight against the barrier of the lips; it hesitates, so to say, and we hear how it clings to the glottis in its slow onward passage. This slight sound, which is not caused by any rhythmic vibration, but only by a certain narrowing of the chordæ, is all that can be meant when some grammarians call these mute consonants sonant; and, physiologically, the only appreciable difference between p and b, t and d, k and g, is that in the former the glottis is wide open, in the latter narrowed, but not so far stretched as to produce musical tones.

1 1. c. p. 9.

Lastly, g, d, b, may be modified to ng, n, m. For these three nasals a full contact takes place, but the

Fig. 26.

breath is stopped, not abruptly as in the tenues, but in the same manner as with the mediæ. At the same time the breathing is emitted, not through the mouth, but through the nose. It is not necessary that breath should be propelled through the nose, as long as the veil is withdrawn

that separates the

Water injected into the

nose while n and m are pronounced rushes at once