I can't provide a code reference, but as a physicist, I can say that whenever grounding critical electronic equipment, whether it be for an alarm system, a stereo system, or a computer network, there is always a choice of many metallic paths that eventually find a route to the earth. Many of these options may not be good ones for the following reasons.
The primary purpose of grounding in electronics is to avoid having the metal box containg the electronic circuits "float" and thus unable to provide shielding from EMF radiation which generates noise in these circuits.
Since the failure of an alarm control circuit often is more critical than a failure of a computer network circuit, choosing the best route to earth ground is important.
The only way to be confidant that the ground wire from the alarm panel connected to the water pipe stands the test of time, is to bring that ground wire (without splices) directly to within 5 feet of the water entrance, assuming that this same water pipe is buried in the soil outside the building.
I personally have more confidence in the ground wire connecting the electrical panel to the copper rod which is driven into the soil, but I think the alarm manufacturer wants to minimize the cost of grounding by allowing a water pipe within the building to be substituted.
The alarm manufacturer is assuming that no segment of the chosen metal water pipe will ever be altered (by introduction of a threaded union with plumber's sealant) or replaced in part by a plastic pipe.
Almost as important, is the problem in electronics due to infamous "ground loops". This is a situation where multiple pieces of electronic equipment which are all interconnected to each other are also individually tied to the electrical ground of the building via different paths.
What happens is that even though each piece of equipment is grounded, the ground wire for each piece of electronic equipment can have a different distance (and possibly a different route) back to the main panel.
The difference in these ground paths causes a small (millivolts) amount of 60 hertz electrical voltage to be picked up within the ground wires which are acting as antennae. All the grounded electrical boxes are pulled to this same voltage(s).
Normally a few millivolts of voltage in ground wires wouldn't bother anybody. But for stereo equipment where the interconnections are often over unbalanced RCA cables and where the audible signals are measured in microvolts, we often hear the background hum in our loadspeakers. When this happens, it is necessary to ensure that only one piece of equipment (usually the receiver) is officially grounded and all the other equipment is "floating" by blocking the ground pin in the 3 pin plugs of all equipment except the audio receiver. Sometimes we even need to "float" the receiver to mimimize the hum.
For TTL and CMOS logic circuits such as used in computer networks or alarm control, the operating voltages are often between 2 and 5 volts DC, so if we have a ground loop generating a few millivolts in the alarm box, the control circuits have adequate immunity and the background hum will not interfere with the circuitry.
However, and this applies for all sensitive electronic equipment that is grounded, if there is a major electrostatic discharge within or nearby the building (lightning, high amperage short circuit, etc), the ground wires will pick up the pulse which in this case may be substantially higher than the few millivolts from ground loops, and there is a risk that our electronics will be fried. The use of varactors (as found in power bars) will provide the "surge protection".
But starting with a robust copper ground wire between the alarm equipment and the point where the water pipe enters the building and keeping this wire as thick as possible and its length as short as possible, will protect the alarm equipment in the best possible way.
As always, the ground connection to the water pipe has to be as professional as the ground connection at the alarm box, that is, solid connectors and no possiblity of corrosion ever reducing the electrical contact.
[This message has been edited by montreal (edited 03-21-2006).]
