Detection of tau-neutrino in DONUT experiment

Question

I try to grasp the significance of the DONUT experiment (Run1: arXiv:hep-ex/0012035, Run1+Run2: arXiv:0711.0728 ). In the case of µ-neutrinos ~100% muon-neutrino produce ~100% muon, no electron. In the case of DONUT the result seems to be considerably more ambiguous. Run1 yields 203 total- / 4 tauon- (2%) neutrino events, run2 375 total- /5 tauon-(1.3%) events. Background is given with 0.34 events. However, the total error of correlation tau-input - tau-output seems to be more like a factor of 2 judging from the change in ratio tauon/all neutrino events from run1 to run2 as well as the change in calculated tau-neutrino-input percentage from 5% to 3% between both papers. This seems to be a rather rough correlation and no conclusive proof for a unique relation between tau-neutrinos and tauon production.

2 Questions:
1.) did I get something wrong ?
2.) is this cross-checked with corresponding experiments probing tauon production from electron + muon neutrinos only (which I assume requires appropriate detectors and is not a byproduct of just any neutrino experiment).

This is only about this particular experiment, not about 3 generations etc.

Answer 1

I believe that for Run 1, significance Z=4/sqrt(0.34)=6.9 sigma > 5 sigma, so "observation" You may use Asimov formula instead of S/sqrt(B) to be more precise.

For Run 1+Run 2, Z=9/sqrt(1.26+0.22)=7.4 sigma

Everything seems fine. For your comment on large variation between Run 1 and Run 2, you can't say that because there is an important uncertainty on low numbers. So you can't conclude that there is a large change : to conclude it, you would need to consider the uncertainty on the the numbers 4 and 9 (sqrt(4) and sqrt(9).

But the significance (Signal / sqrt(background) is transparent for that).