in fact, regarded as indispensable wherever truth is required, yet that of scraping is calculated to produce a higher degree of truth than has ever been attained by grinding. In reference to both processes a great degree of misconception prevails, the effect of which is materially to retard the progress of improvement, and which it is of great importance to remove. While grinding is universally regarded as indispensable to a finished surface, it is, in fact, positively detri-
pair of true surface plates. If one of them be allowed to slide on the other so as to exclude the air, the two plates are caused to adhere together with considerable force, by the pressure of the atmosphere. The surfaces should be well rubbed previously, with a dry cloth, till they are perfectly free from moisture, that the experiment may afford a fair test of accuracy. If any moisture be present it will act like glue, and cause adhesion to take place, supposing the surfaces to be much inferior. But if they be perfectly dry, adhesion proves a high degree of truth, rarely attained.
The experiment may be varied, by letting one surface descend slowly on the other, and thus allowing a stratum of air to form between them. Before they come into contact, the upper plate will become buoyant, and will float on the air without support from the hand. This remarkable effect would seem to depend on the close approximation of the two surfaces at all points, without contact in any—a condition which could not be obtained without extreme accuracy in both. The escape of the remaining portion of air is retarded by friction against the surfaces, the force of which nearly balances the pressure of the upper plate. If one end of the upper plate be slightly raised, and allowed to fall suddenly, the intervening air will act like a ushion, causing a muffled sound to be emitted, quite different from that usually produced by the concussion of metallic bodies.
