Your first paragraph demonstrates the need for the "before" test. Your knowledge of where swarms have arrived may well subconsciously affect the "dowsing response". This is one of the types of problem the scientific method is intended to compensate for.
I don't design experiments for a living and it's a genuinely difficult thing to do well, but I'd suggest you'd need to do something along these lines:
Pick a number of unfamiliar locations large enough to site a good number of bait hives and dowse them to map the "energy lines". Preferably have multiple unconnected people dowse them independently, because if the maps don't align significantly then you're already in trouble. Have someone else who has no knowledge of the map randomly place the bait hives (which must be as near identical as possible) randomly around the locations, but including the positions indicated by whatever intersections of the "energy lines" exist. In all respects other than their proximity to the "energy lines" they should be as indistinguishable as possible. Monitor the hives over the course of a season to see which are selected by swarms, and replace each used hive with a fresh one each time it is selected. Repeat the experiment over a number of years, switching the individual hives between locations at random.
Once the experiment is complete you can tally the results and see if there's any apparent preference for the hives at the intersections. Then consult a competent statistician to see how likely the result might be to have occurred at random.
There are a number of issues you'd have to decide beforehand, such as "how wide" an "energy line" or the area of intersection might be, how you can validly account for any disparity in the "energy line" maps, how you're going to obtain your swarms or if you're going to let them turn up at random. The moving of hives may ideally need to be decided beforehand. I don't know how many hives you'd need at each location, nor how many locations might be desirable. A statistician would probably need to advise. You'd need to decide if all bait hives will be the same size, in the same orientation and so on. All decisions about how the results will be analysed must be made before the experiment commences. And you have to accept that you will have no knowledge of which intersection/non-intersection hive has which swarm count until the experiment is over so that (for example) any decisions you may need to make or information you have to give to other people involved in the experiment cannot be distorted by what you know about the results so far.
Even then, there's clearly no guarantee that "energy lines" exist nor that bees or dowsing can sense them because there may be some other aspect of the landscape to which the dowsers and bees are responding, but even demonstrating the ability to reliably predict the preferences of a swarm would, I'd guess, be a very interesting result and warrant further investigation.
This may not be sufficient, but it's my best guess for the moment. As I said, I don't do this for a living, though I used to know people who did.
James
