the University of Strassburg, where I was studying as PARKER Fellow of Harvard University. While demonstrating with methylene blue the innervation of the frog's heart, I obtained several interesting preparations of the nervous elements in the palate, which led me to a further investigation of their structure. This investigation has enabled me not only to verify several points which have been hitherto in doubt, but also to observe new structures which other investigators have either overlooked or have failed to demonstrate.

In the present paper I shall first give evidence from preparations of the normal palate to show that the fibers of the networks described are true nervous structures. And, secondly, from degeneration preparations I shall endeavor to show whether the cells present in these networks are sheath cells or are as truly nerve cells as those of the brain and sensory ganglia.

I. THE NERVOUS STRUCTURES OF THE PALATE.

Preparations were obtained by injecting 1⁄2 cc. of a 1% solution of methylene blue (in normal salt solution) into the abdominal vein of the frog. The animals were either rendered passive by the subcutaneous injection of curare, or tied out immovable on the wooden frame shown in figure 8, p. 107. Within five or ten minutes after the appearance of the stain in the integument, the palate with its nerves and vessels was dissected from the roof of the mouth-an easy task, thanks to the lymphsinus lying beneath the integument. The preparation was then placed epithelial side down, in a flat watch crystal and the exposed surface moistened with the animal's blood while the progress of the stain was watched under the microscope. When the right degree of staining was judged to be obtained, the blood and mucus were rinsed away with normal salt solution and the tissue fixed with ammonium picrate. The preparations were first usually mounted in glycerin, studied in the fresh condition, and important details sketched with the camera lucida. They could then be quickly washed in water, refixed in ammonium molybdate and mounted in balsam. The molybdate method gives much clearer mounts, but has this disadvantage, that the

finer details are often lost by the washing out of the stain in running the preparations up through the alcohols. By studying preparations by both methods I thus did away with the disadvantages of each. Many mounts were made between two cover glasses, allowing the use of an oil immersion from both sides. This is a distinct advantage when whole mounts are employed.

The frog's palate is innervated chiefly if not exclusively by the Ramus palatinus of the seventh cranial nerve (facialis). Each palatine branch (Fig. 1) passes down to the roof of the

Fig. 1. The roof of the frog's mouth with the integument dissected away to show the course of the palatine nerves. R. palatinus, palatine nerve; R. comm. V. 2, Ramus communicans of the trigeminal nerve; a, b, points at which nerve was severed. (After GAUPP).

mouth immediately anterior to the lateral process of the basi sphenoid, runs nearly straight cephalad and then, bending sharply laterad, joins the Ramus communicans of the trigeminus. Along its course the nerve gives off many lateral twigs, the fibers of which interlace to form an intricate plexus of medul

lated fiber-bundles over the whole inner surface of the palate. From this plexus fibers pass off to the epithelium.

According

to BETHE each medullated fiber divides into four branches and each branch innervates a different sensory organ, the number for each organ being but two. This peculiar and definite method of innervation, he points out, is the most natural arrangement by which each end organ may receive a distinct nerve supply at the expense of the smallest number of nerve fibers a separate nerve supply for each sensory organ being requisite for the localization of tactile stimuli.

My preparations confirm the general conclusions of BETHE, but the distribution of the sensory fibers is not as simple as he supposed. It is true that usually only two or three large fibers innervate each sensory organ; these break up into numerous fine fibrils, which, after a tortuous course, end between the epithelial cells. The sensory organs in which these fibers end project slightly above the surface of the palate and are most numerous at the sides. BETHE Counted an average of 210 end organs but only 70 fibers in the palatine nerve; if each fiber branches into four, as BETHE asserts, this would allow an average of between two and three branches for each sensory spot. But in addition to these branches I find numerous bundles of fibrillae given off from each medullated fiber. These divide into still smaller fibrils which form a network of fine neurofibrillae and probably connect the different sensory organs. This network has not to my knowledge been observed in the integument of the frog, but SFAMENI (:02) and RUFFINI (:01) have recently described structures apparently identical to it in the skin of man. In the frog the fibrillae composing the network are very difficult of demonstration. In the great majority of methylene blue preparations they are but incompletely stained, and of good preparations I obtained but two or three out of perhaps a hundred trials. The network lies directly beneath the epithelium and is composed entirely of non-medullated fibrillae (Fig. 2). Strands of these are given off from the medullated fibers as seen in the figure at a and a'. The strands divide and their fibrils are apparently continuous with each other in a fine

meshed irregular network; for under a high magnification some of the meshes were found to be formed of a single fibrilla, and it is absolutely impossible to say where one fiber ends and another begins. Certain fibrils from this network end in the sensory spots of the epithelium (Fig. 2, b, b'); others terminate freely in the regions between the sensory organs. There

is thus a diffuse sensory nerve supply throughout the epithelium of the palate.

From certain preparations in which the fibrils of this network were incompletely stained, I was able to trace a single fibrilla from one medullated fiber into another without loss of continuity (Fig. 3). This connecting fibril appears to be homogeneous in structure throughout its entire length. It is very possible that such a condition may be produced by the overlapping of two fibrils. I myself have concluded after a careful study of these networks that they are formed by such an overlapping of two fibrils and not by the direct union of fibrils from different "neurones." In either case it is impossible to say that the fibril belongs to one neurone or to the other.

The presence of fibrillar networks throughout the integument is what we should expect from the physiological facts as to tactile sensations. The whole surface of the skin is more or less sensitive to tactile stimuli but the localization and acuteness of the sensations depend upon the presence of special receptive organs. If two stimuli are applied to a region between sensory spots only one sensation is felt, as the stimulus is diffused and affects equally a number of neurones. If, however,

the two stimuli are so far apart as to affect different sensory spots, innervated by distinct nerve fibers, each of these will be strongly stimulated, and two distinct sensations will be the result.

Fig. 2. A subepithelial network of neurofibrillae from the palate of the frog; a, a', strands of fibrillae from medullated fibers; b, b', fibrillae which apparently end in the sensory spots; c, c', two sensory spots. The network is viewed from the epithelial side of the palate, and the terminations of the sensory fibrils among the cells of the epithelium are indicated by knob-like enlargements. X 330; details with LEITZ 1-12 oil immersion.