WOW! That's a big difference!
I've read that the Apple assemblers get a little sloppy with the thermal paste and often use too much, which, as you know, has the opposite effect of its intended use. May I ask which brand of thermal paste you used? I was looking at the Arctic 5, but a lot of people are switching to Arctic MX-4 because it's easier to apply and is non-conductive.
Also, what type of application did you use—pea sized dot, line, x-pattern, etc.?
Tiny pea-sized glob, that I then spread around with a zip-lock sandwich bag that I wrap around my index finger. Gives me a smooth plastic surface around the finger tip.
Keep in mind, paste does have a cure time. The time to cure varies with brand as well as type. Artic Silver 5, for example (what I used with my MBP) takes up to 200 hours to cure. So, temps shouldn't be measured until the curing is done.
There are lots of theories around using lines, or drops, or spreading. I've tried them all, and never saw much of a difference worth noting between any of them. The key with all of them is that there needs to be paste coverage across the cores of the processor. So I just play it safe and spread the crap over it.
My dual socket Xeon host had pea drops done to it when I built it. Idle temps were 28-30C (X5650's.) Next time I did them, I used lines, and got same results. Third and last time I did it, I just spread it like I did with my MBP, and got 29-31C. But a 1 degree diff could be attributed to me using different paste the last time, because I ran out of the Noctua paste that came with the heat sinks and fans. So, I'm not worried about it for now. Next time I do it, I'll try to get my hands on some more Noctua.