Interpreting B and C for cmor wavelet

[Eli.Ate...@water.ca.gov]

I have a series (time units are hours, 4 samples per hour). My signal is tidal, so it has known periods at 25,12.5,8,6.25 hours as well other more vague overtides and subtide components respectively at higher and lower frequencies that to first approximation subdivide the lunar day. 

I am trying to construct a cmor wavelet but I'm getting a bit confused about B and C. I'm using hours as the underlying unit so my sampling_period is 0.25 hours. My understanding is that the default wavelet range is [-8,8] data points, so that width would be about a total of 4 hours, right? How is the parameter C related to this. Is it in my units or more generic ones? It is pretty short compared to the  finest frequency of interest to me would have a period of 4-6 hours, so it seems like I need to use scale cleverly? I'm ideally interested in a narrow enough bandwidth to separate 8 and 6.25 hour periods or at least the 12.5 and 6.25 hour periods if reconstruction is an issue. It is well known that the time-frequency limits are harsh with nonstationary tides.

Can someone help me map these units to what I should be using in pywavelets? Obviously I'm just providing one central frequency plus a scale, but I'm not sure what that one central frequency should be or of its units. Thanks!