Millimeter-wave (mmWave) communications promise Gigabit/s data rates thanks to the availability of large swaths of bandwidth between 10-100 GHz. Although cellular operators prefer the lower portions of the spectrum due to popular belief that propagation there is more favorable, the measurement campaigns to confirm this -conducted by ten organizations thus far -report conflicting results. Yet it is not clear whether the conflict can be attributed to the channel itself -measured in different environments and at different center frequencies -or to the differences in the organizations' channel sounders and sounding techniques. In this paper, we propose a methodology to measure mmWave frequency dependence, using the 26.5-40 GHz band as an example. The methodology emphasizes calibration of the equipment so that the measurement results represent the channel alone (and not the channel coupled with the channel sounder). Our results confirm that free-space propagation is indeed frequency invariant -a well understood phenomena but to our knowledge reported nowhere else at mmWave to date. More interestingly, we found that specular paths -the strongest after the line-of-sight path and so pivotal to maintaining connectivity during blockage -are the least invariant compared to weaker diffracted and diffuse paths.