In 2012 I had the opportunity to discuss Henriot's paper on arithmetic progressions in A+B+C with him. Out of these discussions came a slightly different approach to Roth's theorem. Of course now Bloom has produced a new proof of Roth's theorem by developing his wonderful strengthening of Chang's theorem. At the time the best know approach required the Croot-Sisask Lemma as well as an application of the Dyson e-transform. From our conversations we managed to remove this appeal to the Dyson e-transform and I shall discuss this here.