I would propose that the question is not "is there any refraction in any given water convexity experiment?", rather it's "how much refraction is there in any given water convexity observation?" It is well known that air and water temperatures, humidity and air pressure can vary significantly and affect the refractive properties of the air near the water surface to unknown degrees and thereby affect the reliability of such observations. I would therefore propose that all water convexity experiments (on both sides) be deemed inconclusive, and therefore inadmissible, unless the atmospheric and sea conditions can be documented along the entire length of the observed distance.