Continuing these experiments with most fervent zeal, taking no account of the passage of time, with an utter disregard for meals, and but scanty hours of sleep snatched reluctantly at odd periods of the day or night, Edison kept his laboratory going without cessation. A great variety of lamps was made of the platinum-iridium type, mostly with thermal devices to regulate the temperature of the burner and prevent its being melted by an excess of current. The study of apparatus for obtaining more perfect vacua was unceasingly carried on, for Edison realized that in this there lay a potent factor of ultimate success. About August he had obtained a pump that would produce a vacuum up to about the one-hundred-thousandth part of an atmosphere, and some time during the next month, or beginning of October, had obtained one that would produce a vacuum up to the one-millionth part of an atmosphere. It must be remembered that the conditions necessary for MAINTAINING this high vacuum were only made possible by his invention of the one-piece all-glass globe, in which all the joints were hermetically sealed during its manufacture into a lamp, whereby a high vacuum could be retained continuously for any length of time. In obtaining this perfection of vacuum apparatus, Edison realized that he was approaching much nearer to a solution of the problem. In his experiments with the platinum-iridium lamps, he had been working all the time toward the proposition of high resistance and small radiating surface, until he had made a lamp having thirty feet of fine platinum wire wound upon a small bobbin of infusible material; but the desired economy, simplicity, and durability were not obtained in this manner, although at all times the burner was maintained at a critically high temperature. After attaining a high degree of perfection with these lamps, he recognized their impracticable character, and his mind reverted to the opinion he had formed in his early experiments two years before--viz., that carbon had the requisite resistance to permit a very simple conductor to accomplish the object if it could be used in the form of a hair-like "filament," provided the filament itself could be made sufficiently homogeneous. As we have already seen, he could not use carbon successfully in his earlier experiments, for the strips of carbon he then employed, although they were much larger than "filaments," would not stand, but were consumed in a few minutes under the imperfect conditions then at his command.