Our recent work with the glass electrode has been directed towards the reinvestigation of some of the accepted facts about glass electrode behaviour and performance, which we believe to be insufficiently firmly established, but which nevertheless have been perpetuated by continual repetition in recent monographs. Many of these facts stem from investigations of thirty years ago and have not been checked with modern techniques and instrumentation.
In work to be reported in detail elsewhere a study has been made of the time response and behaviour of representative commercial hydrogen-responsive glass electrodes in acid 1'2 and in alkaline solutions 3. During the course of this and other studies small errors of 0.005-0.02 pH were observed in certain buffer solutions where the concentration of the buffer constituents was low 4. Recently one of us has been concerned with the study of glass electrodes in heavy water solutions 5.
In this paper we have made a study of the effect of varying the inner fillingof the glass electrode. It has always been assumed that apart from (a) determining the "null-punkt" of the electrode and hence its suitability for use with certain pH meters, and (b) allowing temperature effects to be reduced, the only property of the internal reference electrode system was to provide a constant, or at least only slowly varying, potential 6. On the other hand, there have been reports recently 7 -10 that, for potentiometric titrations in partially aqueous or (almost) non-aqueous media, it was advantageous to use the same solvent medium on the inside of the electrode as on the outside. For example, claimed that an acetonitrile-filled electrode gave more reproducible and more rapidly attained potentials for titrations in that solvent. Douheret8 also claimed faster response and more precise determinations for methanol media. Similar statements have been made for titrations in dimethylsulphoxide 9'1ยฐ and dimethylformamide 1 ~. It was suggested many years ago that mercury was quite adequate as an internal filling x2 and this was confirmed for the titrations in dime-