Luminous bacteria are so small that the light from a single individual cannot be seen. It is almost impossible to make out structural differences within the cell and we cannot definitely state in just what special region, if any, the luminescence is produced. We do know that the light is intracellular and that filtration of the bacteria from their culture medium gives a dark sterile filtrate absolutely free from any luminous secretion. Among protozoa, in certain forms at least, it is easy to observe that luminescence is connected with globules or granules which were considered by the earlier observers to be oil droplets. Thus, in _Noctiluca_ (Figs. 17 and 18), when the animal is violently stimulated or in the presence of reagents which slowly kill it, the whole interior appears a mass of starry points of light which can be traced to minute granules along the strands of protoplasm (Quatrefages, 1850). [Illustration: FIG. 17.--_Noctiluca miliaris_, showing photogenic granules in cytoplasm. _n_, nucleus; _c_, cytoplasmic strands containing photogenic (large) and other (small) granules; _p_, pharynx; _f_, flagellum; _o_, oral groove; _t_, tentacle; _s_, spines at base of tentacle; _v_, vacuoles. _Drawn by E. B. Harvey._] [Illustration: FIG. 18.--_Noctiluca miliaris_ as it appears during luminescence (_after Quatrefages_). Upper left and middle, low power; below, high power; upper right, a crushed fragment still luminescent.] Turning to the multicellular forms, we find the simplest development of luminosity in those animals which possess gland cells producing a luminous secretion. These cells may be scattered over the surface of the animal as in _Chætopterus_ (Fig. 19) or _Cavernularia_, or restricted to certain areas [_Pholas_, (Fig. 19),] or more definitely localized to form an isolated group of gland cells as in _Cypridina_. True multicellular glands also occur. In every case, however, we find that the luminosity of these uni- or multicellular glands is connected with the presence of granules. They are often spoken of as _luciferine granules_, although it is not certain whether they are made up of luciferin or luciferase (see Chapter IV) or both. They are most similar to the zymogen granules found so abundantly in gland cells and thought to be the precursors of various enzymes. According to Dahlgren (1915), the luciferine granules stain blue-black by iron hæmatoxylon after fixation at the boiling point, and photogenic cells can be detected by this method of selective staining. Dubois (1914, book), who regards them as examples of _bioprotein_, comparable to the chondriosomes and handed on from one generation to another, gives them the name of _vacuolides_ or _macrozymases_. In some forms he has described their transformation into crystals and believed at one time that animal light was a crystalloluminescence. His figures of the crystal transformation are not very convincing. Pierantoni (1915) has considered the granules to be