The limited range of previous measurements of thermal agitation voltage is pointed out and data are given which extend the range of resistances and frequencies considerably. The results substantiate Nyquist's theory for resistances up to 1.7×109ohms and for frequencies between 1 and 10,000 cycles per second. A value of 1.366×10−16ergs per degree with a probable error of 0.016×10−16ergs per degree is obtained for Boltzmann's constant. This is nearer to the accepted value than previous determinations by this method have given.