I don't think that the alternator will have that large an effect on gas mileage, maybe someone on here knows the calculation for converting watts to petrol used, I've seen it ages back in a study of how much extra fuel would be used nationally if everyone drove with their lights on all the time & vaguely remember it was in the under 10% region. However, maybe using the battery only you are babying it along at under 5000 rpm, while with the alternator connected you're hitting the revs much harder... that'll make a big difference in mileage.
I thought that'd be an issue, also, but I ran quite a few controlled runs around over a set distance through the same ammount of fuel and it all came out about the same.
When I first fixed the alternator, I definately was revving it a bit tighter, simply because I couldn't do that before! But after the first tank of getting 35mpg I slowed way down. I ran the controlled tests, ran 3 tanks with the alt connected and 3 with it disconnected. Driving style the same through all 6 tanks. The only difference was the alternator. With the alt connected, 40 mpg average and with it disconnected, 57mpg average. This was a pretty controlled test (I've been testing MPG mods for years).
I'll have to add the watts consumed by all of the lights to convert it to horsepower.
1HP = 746w.
The entire charging system is only about 250 watts, I believe, so it should pull about 3 horsepower. Given any innefficiencies in the system, I'll be generous and say 4 hp.
If my bike produces 50hp, than 4/50=8%
At 3500rpm (where I normally drive) it doesn't have anywhere near 50hp, probably more like 10 or 12. If it's 12hp, than 4/12=33%.
If at 3500 rpm and 5th gear I get 57mpg, if I lose 33% of that I should be getting 37.62 mpg, or about 40!
Wow, That should prove it. I guess all I can do to try to combat this effect is to minimize any electrical load I have on the system.