3. I found that if the emulsion be rubbed (so to speak) on to the glass there is much less chance of frilling, etc., than if it were poured on. I think it is because in the former case the gelatine is in firmer contact with the glass. When the gelatine is poured on to the plate the cold glass instantly chills it, and by the time the emulsion has reached the edges of the plate it has so far set as to have partially lost its power of adhesion to the smooth surface of the glass. [Illustration: Fig. 1.--Showing melted emulsion in coater ready for coating.] Two or three years ago, when it was the practice to warm the plates before coating, I found from a series of experiments I then made that when a plate was warmed before being coated the emulsion commenced setting on the surface of the film, and of course in setting contracted, thereby leaving a partial vacuum between the film and the glass. On development frilling was the consequence. I found, however, that, when pouring the same emulsion on cold glass, on the portion of the plate where it was poured on, the film instantly chilled and commenced to contract on to the glass, and it never frilled there; but toward the edges of the plate, as the emulsion had commenced to chill before they were covered, the film was not in such perfect contact with the glass. Any person can try the experiment by first coating a plate in the ordinary way, and on the second plate just pour a small pool of emulsion on the center; let both dry, and he will then see after exposure which frills the easier on development. [Illustration: Fig. 2.--Showing emulsion flowing through the slit on to the glass.] After a series of experiments I found that by brushing a substratum of emulsion on to the cold plate (with a brush made by binding a strip of wash-leather at the end of a strip of glass), and then pouring the full quantity of emulsion on to the substratum (for quarter-plates I used a small silver teaspoon, which held sufficient to cover that size of plate), I found I could coat plates far better and quicker and as easily as when coating with collodion, and I got over the difficulties of having frilling plates. When only a few small plates are required--such as for experimental purposes--I believe this method is as quick and good as any; but when several dozen plates are wanted, any plan of coating them separately takes a long time. With my plate-coater I can coat a dozen plates in about the time I formerly took to coat one. When coating a number, I thought it would be best to lay them in rows on the leveling-shelves and draw the receptacle containing the emulsion over them, rather than keep the latter a fixture and run the plates under it either on an endless band or sliding shelves; because by the first mode the plates can be fixed close together, and the emulsion is less likely to get between them.