Basically, my supervisor wanted to measure variable X. X is very important to him due to some other projects we are working on. So he wanted me to give a talk on measuring X at an upcoming conference.
I was skeptical -- I thought that maybe X wasn't a real variable, but just a mathematical effect that would cancel out if we did our measurements right. My supervisor gave me some compelling arguments and convinced me that X was real and we should measure it, so I submitted an abstract to the conference and got accepted.
Now, having done the work and taken the data, I am once more convinced of my old idea that X isn't a real physical thing. The talk is coming up soon and despite the enthusiasm that appears in my abstract the best I can honestly say is that we took some measurements and got no results. I know my supervisor would be upset if I give a talk to the effect of "X doesn't matter" because that reflects negatively on his other projects. Should I cancel the talk, change the content so it's drastically different than the abstract, or something else?