To summarise the method of bulk summing the spectra, I've created frequency vs time spectra by using the e143_analyser.py script to stack a series of -s slices. I applied it to some of last runs data with 72Ge, to the 72Br data, last night's less-scraped 70Se, and tonight's more-scraped 70Se, respectively.

A few comments: the 72Ge isomer trail is obvious, but I would never have found it in this total sum if I didn't know where to look. I had to turn up the contrast a lot before it was visible. The 72Br isomer is less obvious than in the single particle tracks, but it too is obvious. You can see that the peak is wider to left at the injection, and 
then narrows out as the isomer decays. This is what we would see if the two peak weren't well resolved.
With that in mind, the 70Se data is challenging. We've established the broad peak is not the isomer as it is not present in the 245MHz detector., and the many single lines seen in the well-scraped data correspond in frequency exactly to this broad band. The 70Se peak doesn't get broader at the base at injection like 72Br, and there isn't any 
line visible in the sum in between. I was nervous that the contrast might be the problem like with 72Ge, but I turned up the contrast on the less-scraped data, but no signal emerged.

So I conclude that if the 70Se isomer is to be found, it needs to be done on a shot-by-shot basis with comparison to the sum to see if it matches the intruder broad peak.