If you think about a simple 360° sine wave (the traditional AC wave form, positive 180° first and then rolling through 180° negative), the waveform is only actually peaking for a fraction (technically we're into calculus at this point) of a moment. Leaving aside the complications of RMS [Root Mean Square] values for now, what we measure using the unit of Volt(s) is the difference between two points - Potential Difference. So for the entire time our sine wave is swooping along, it's reference point is both mathematical and practical zero - or to put it another way, it's neutral to either the +ve or -ve phase of the wave. By then connecting that neutral to all the other neutrals around us (which is the principle of earthing) we know that there is no [Potential] difference between that neutral and any other so by doing so we sort of calibrate our live waveform - ie, if you were to take a probe on a meter to the live on one house, and the other probe to the live on another house you should (theoretically) have a reading of 0v - because there is no potential difference between them.
On a three phase system, things change again because now we have 3 lots of sine wave all interpolating at 120°, so when looked at simply (again, ignoring RMS..) you near as damn it always have something at the top, and something at the bottom, so the potential difference there is (simplified..) +230 to -230 which gives us 460V (it's actually lower) and mathematically nothing is ever zero because as soon as one waveform hits zero the one behind it is roughly 160v ahead of it already, followed by another. This is the reason (well, one of a few) why we don't even need a neutral to make a true 3ph circuit work.