Here's some math on the analysis. This is based on a formula posted by Aldriana (my statistics skills are extremely rusty, so I'm going to have to trust the formula atm).
Formula: Standard Deviation of an Event = sqrt(event * (samples - event) / samples) / samples
So for the 2.6 speed weapons case:
sqrt(28 * (946 - 28)/946) / 946
0.551% Std Dev
Average is (28 / 946) = 2.960% proc rate
Since during independent event testing, we can be 95% confident that the sampled results will fall within 2 standard deviations of the mean, we can be 95% certain that the proc rate for 2.6 speed weapons is 2.960% +/- (2*0.551%) = 1.854% to 4.062%
(We can also be 68% certain that it falls within 1 Standard Deviation, namely 2.409% to 3.511%)
For the 1.5 speed weapons case:
sqrt(42 * (1375 - 42)/1375) / 1375
0.464% Std Dev
Average is (42 / 1375) = 3.054% proc rate
95% certain that proc rate for 1.5 speed weapons is 3.054% +/- (2*0.464%) = 2.126% to
3.952%
68% certain the proc rate is 2.590% to 3.518%
If Romulo's was a PPM trinket, then we would expect the 2.6 speed weapon to proc more than the 1.5 speed weapon. Namely, we'd expect (2.6/1.5) = 1.73, or 73% more procs per swing.
Using the highest numbers from the 2.6 speed test and the lowest numbers from the 1.5 speed test we get...
95% confidence: 4.062% / 2.126% = 1.911 = 91.1% more procs
68% confidence: 3.511% / 2.590% = 1.356 = 35.5% more procs
So we can NOT be 95% sure that it's not a PPM, but we can be at least 68% sure. I don't have the time at the moment, but we can actually figure out exactly how sure we are (somewhere in the 80-90% range likely).
1 final calculation before I go, assuming that is NOT a PPM, then we can use all the data together to calc proc rate.
sqrt(70 * (2321 - 70)/2321) / 2321
0.355% Std Dev
Average is (70 / 2321) = 3.016%
95% Likely proc rate for vial is between 3.016% +/- (2 * 0.355%) = 2.306% to 3.726%