The paper describes measurements of surge voltages made under actual working conditions on 33-kV transmission lines both overhead, underground and composite.The apparatus used for the measurements was the klydonograph, and the paper gives a description of this instrument in its final form and a discussion of its working principles. The specially designed auxiliary condensers required to subdivide the line voltage are also described.Measurements on three important transmission systems covering a year or more of continuous recording are detailed. Generally there is a large amount of oscillatory surging set up by brushing at insulators when such insulators are in a dirty condition, the over-voltage being of very high frequency and of the order of three times the normal line voltage.The atmospheric conditions under which salt is deposited on insulators are described and some of the records show the effect on the line voltage of a salt-laden insulator up to the time at which flash-over occurred and the line was put out of operation.The question of porcelain insulators to meet the prevailing foul conditions and to avoid surges due to brushing is discussed. It is shown that with proper attention to design the voltage at which brushing commences can be materially increased, and it is suggested that this factor is of even greater importance than a very high initial flash-over voltage.