I just did the following experiment:
Using a metal Whitman's chocolate box, I placed my phone it it, closed the lid then called it from another phone. To my surprise, it rang.
The normal parameters of a faraday cage is any gaps have to be small compared to the wavelength of the radiation. The edges of the box overlap by about 1 cm, and the gap is a fraction of a mm.
Phones operate at 1-2 GHz. Skin depth for 1 GHz is about 2 microns Wavelength of 1 GHz is about 8 inches.
Even if it's 30 ga metal that's still 12 thou or about 300 microns. Now it's steel, not copper, but steel isn't THAT much worse a conductor.
I don't understand what is going on.
Edit: More experiments:
I have phoning over wifi enabled, and I'm sitting 20 feet from the access point. Disabling the wifi on the phone and using the 2 bar local cell signal, no calls.
Re-enable wi-fi and I can reach the phone in the box intermittently. I can ping the phone from my local network which makes for faster tests. Curiously, the phone doesn't respond to pings when it's asleep.
My current working hypothesis is that the electrical connection between the box and the lid isn't reliable.
The box is laquered with red and gold Christmas colours which may be acting as an insulator.
After opening and closing the box several times, while running a ping to the device over my local I get erratic results. Sometimes closing the box gets me 'request timed out.' sometimes not.
Wrapping it with alumininum foil loosely gives more odd results.
At one point I had it loose enough that I could see the phone -- gap of maybe 1 cm, and got device unreachable, and didn't become reachable until it was open to about 2 inches. I suspect that my AP has some kind of adaptive power setting.
Right now it's sitting in it's metal box and I'll get a series of 20 unreachables, 4 pings with variable times, more timeouts.