Wow, interesting what happens when you don't keep an eye on the forum for a while! I guess I've been quoted a bit.
I'm sorry I can't cite any references but my own on this, I did have enough material together to do an AES paper on it at one time, seems sort of pointless now.
Looking over the posts in this thread, I might be able to shed more light.
First issue: to demag regularly or not. I am not in favor of blind, routine demagnetizing. All I'm really saying is, if the symptoms dictate it, look for the root cause of the magnetism. There's nothing inherent in tape recording or the materials used that should magnetize heads or parts. If you see symptoms pop up, find the cause, because demagnetizing might "fix" it for you, but it's only a band-aid. If symptoms don't dictate it, you can just as easily create a temporary residual field with a degmagger as get rid of it. I'd leave it in the tool box, unless you have an emergency, and even then, only use while monitoring the resulting symptoms. Keep in mind, the record and erase heads have bias and erase current on them when in use, which is a much hotter demagnetizing field that you can induce even with a demagnetizer.
Next, gauss meters and magnetometers were mentioned. I'm not taking a shot at the R. B. Annis unit, it's accurate in the fields it was calibrated for. But, you can't measure the residual field on a head with it. The field is concentrated in the gap, and you can't get that big thing anywhere near the gap. I've spoken to the manufacturer, it's only accurate in "large cross-section fields". A gauss meter with a very small hall-effect probe can intercept a bit of the residual fields on heads, but not enough for accuracy. The bigger problem with either device is, they are sensitive to the earth's magnetic field. You need to find a position where the device reads a null in the earth's field, then swing your tape machine parts into that position. It's really hard to do. Again, ok on large guides, no good on heads.
The only way to tell if you have a magnetization issue is by checking for symptoms. They are: HF erasure with repeated plays (like 25 to 50), elevated noise over virgin tape noise caused by just running tape through the machine, elevated low frequency noise during record, elevated even-order harmonics during record. The even harmonics indicator is VERY sensitive to residual.
Lots of people have a story about how demagnetizing fixed their machine. Well, sorry to say, it didn't "fix" anything, it just hid the real problem temporarily. Hey, that might just work for you! In my world, I like things to work properly, and not have issues that sneak up on you. I'd fix the machine, not just keep demagnetizing it. There are poorly designed tape machines in the world. Some have parts with a natural magnetic state that is somewhat polarized. Cheap (or very old) heads, for example. Machines that kick a huge pulse through the heads when switching in or out of a mode can be self magnetizing. Power on/off transients could be an issue too. But, again, there' nothing inherent in the materials or process that causes magnetization by just using the machine.
I know every test tape has a warning, and every maintenance manual does to. And I'm not saying you should ignore them. However, everyone with a computer, a sound card, and free spectrum analyzer software has what they need to test for the effects of residual magnetism, and more importantly, verify their demagnetizing efforts are really working. I am quite certain, if you really check for symptoms, you'll either find you don't need to demagnetize, or you'll discover a defect in your machine that you can address.
My information was taken from the results of a series of detailed experiments I conducted on this issue many years ago. I was told by the staff at the studio where I worked as an engineer that I didn't need to demagnetize, and I didn't believe it either. I asked permission to research the issue, and was permitted to do so.
I rented a lab-grade F. W. Bell gauss meter, and used a Tektronix 5L4N spectrum analyzer to do the tests on several different machines. Yes, it was in the late 1970s, or early 1980s, that's all we had. The tape stock used was Ampex 630, 406, and 456. To measure the effects of a residual field, I had create one. I used a permanent magnet as the source, and placed the hall-effect probe at the head surface to determine the field strength that I was introducing. I varied the field strength while observing the symptoms. That established the basic flux levels that had to be present to produce the symptoms.
Next, I tried to deliberately magnetize the machine's heads by using the permanent magnet. I succeeded, and got lots of symptoms. I then tried to demagnetize while monitoring for changes in the symptoms. I found that demagnetizing was not as reliable as I though! In fact, several times I made things worse.
My last experiment was to deliberately magnetize a machine, and leave it that way. In 24 hours, the machine had reverted back to a neutral state without demagnetizing.
At the studio I used the 5L4N spectrum analyzer for tape machine alignment, and routinely checked for symptoms of magnetism. I never found any in the 15 years I worked there. The facility had upwards of 30 1/4" machines.
So, I maintain, routine demagnetizing is unnecessary, and not recommended UNLESS you monitor the results with instrumentation, and if it is necessary, go find the cause of the problem, because it should not be. There is nothing in the recording or playing process, nothing inerrant in the heads and guides that results in magnetized parts. Recording signals produce strong AC fields which tend to degauss, and the flux found on recorded tape (AC fields also) is far too low to magnetize anything.
Residual magnetism is the result of a defect, not the process of magnetic recording.
Anyone and everyone is welcome to disbelieve what I say. I just would encourage you to experiment for yourself first, before sticking to widely accepted dogma.
But don't let me stop anything... please. 
/DA
