After replacing the original grounded 2 Outlet receptacle (installed in 1973) with a new TRSGF20 2 Outlet GFCI receptacle, verifying the Line and Load cabling is correct, testing the GFCI with its TEST button, and verifying that the outlets and the download actually have power and turn off when the TEST button is pressed and the GFCI trips, I wanted to do an actual test using an extension cord and bucket of water.

The procedure:

  1. Fill a bucket with water.

  2. Plug the extension cord into the FGCI outlet.

  3. Standing 2 or 3 feet away, drop the extension cord outlet into the water.

The water should trip the GFCI outlet. If it doesn't, it should have tripped the circuit breaker. Neither happened. One of the related questions in StackExchange said "Soft" water could be the problem. So, after unplugging the extension cord, I added salt to the water (the question mentioned that salt is a great conductor), and repeated the test. No luck. Any suggestions?

  • 1
    Did you use a three prong grounded extension cord?
    – JACK
    Oct 29, 2020 at 16:45
  • 1
    Can you provide testing procedure you used to verify you don't have an open ground. The test button on a GFCI channels current around the current detectors, but still routes current from hot to neutral not using the ground. If your ground is open it will not trip by your test, but your circuit breaker should trip if water is adequately doped. Pure water is an isolater. Oct 29, 2020 at 16:49
  • Even plug in testers do not function with open ground. For instance read the first paragraph of page 5 of this tester images.homedepot-static.com/catalog/pdfImages/8c/… Oct 29, 2020 at 16:56
  • Thanks Jack. I tried both types of extension cords (3 prong grounded and 2 prong ungrounded). As for the Open Ground question, I really don't know how to verify whether or not the ground wires (1 for the Line cables and 1 for the Load cables) are an Open Ground. I can only tell you that, in the circuit breaker box, there are a lot of ground wires. in the circuit breaker linked to the GFCI cabling. The 'Line' ground follows the neutral and hot wires going out of the actual outlet box which is steel. The 'Load' ground wire is actually attached to the outlet steel box. Oct 29, 2020 at 19:44
  • 1
    Why are you so paranoid about the built-in TEST button? That button is required by UL 943 section 5.15.3 to generate what basically is a ground fault (flowing ~10mA from load-hot to line-neutral or vice versa): "The supervisory circuit shall operate to create a current imbalance of the sensing toroid, but shall not rely on the equipment grounding conductor to operate." Oct 29, 2020 at 23:53

4 Answers 4


Don't do this! PLEASE!

First of all, this is just simply not a good idea. It proves very little, if anything, about functionality of the GFCI. In fact, if you have a GFCI with ground not connected, this might not trip at all, as the electricity won't actually flow through the ground wire (since it is not connected to anything - no complete circuit) and therefore all flow back through neutral until someone gets curious and wiggles the wet extension cord and gets themselves to be part of the circuit! It is actually dangerous, particularly if the GFCI, for some reason, does not trip when it should. Don't do this.

As far as the actual question regarding salt. Salt is NOT a good conductor. Pure water is NOT a good conductor. Water with dissolved salt conducts better than pure water, but it still doesn't conduct all that well. The danger with water and ground-faults is that even if the water only conducts a little bit, it conducts far better than air and conducts enough to kill. You may think of "electric chair" type voltage/current to kill, but a very small current at 120V is enough to kill if it goes in one arm and out the other through your heart.

If you don't trust the built-in tester (and not knowing how it works, that is actually a reasonable consideration), get a plug-in tester. The plug-in testers work by diverting a known amount of current to the ground pin within a safe, insulated, device. FYI, the plug-in testers will (like the crazy extension cord bucket test) fail if there is no ground wire connected to the GFCI, but that does not mean the GFCI won't function properly.

While we're at it, don't do this either...

Do not test your smoke detector by burning some paper, instead of relying on those oh-so-unreliable-just-like-a-GFCI built-in test button. While holding that little piece of paper you may drop it and start a fire. And if it doesn't set off the detector because it didn't produce enough smoke of the right particle size, you'll be tempted to make bigger fires until the bonfire gets out of control... And if you do manage to set off the detector with a small, contained, safe, fire, you will inevitably find out that your wired smoke detector is connected to an alarm control panel that automatically calls the fire department while simultaneously activating alarms through the entire high-rise building. Inevitably resulting in a fine for a false alarm and/or an eviction notice.


  • 1
    Thanks for your response Manassehkatz. I'm going to get a plug-in tester. As I told Jack in an earlier answer, there are 2 ground wires (1 from the Line cables and 1 from the Load cables). The 'Line' ground follows the neutral and hot wires going out of the actual outlet box which is steel. The 'Load' ground wire is actually attached to the outlet steel box. Hopefully this means there is no open ground. Thanks again. Oct 29, 2020 at 19:50

Spend $3-10 on an outlet tester with a GFCI test button.

Far cheaper than the extension cord you are ruining with salt water.

This "test" smells strongly of "stupid internet tricks."



Yes as other have stated DONT do this with live house power. If your interested in what the restively of your tub water is, UnPLUG the cord, drop the end in the water and measure the resistance across the HOT and neutral plug wires (WITH IT UNPLUGGED). the Hot to ground is likely a similar value as the distances between the bare parts of the extension plug are similar, all about 1cm. The resistivity of water is very corelated to the ions present. Pure water has resistivity of 100,000 ohms-m while tap water will be more in the range of 5 to 500 ohm-m.

But what does all this mean?

So that tap water at 1cm between the electrodes results in about 500 to 50,000 ohms initial resistance which is what this experiment should have measured.

But 10s of thousands of ohms doesn't sound scary? 10,000 ohms at 120 V will only flow 12ma?

12 ma will shock you, but it would need to be an order of magnitude higher to be really life endangering. HOWEVER, as the electricity heats the water up between the conductors, its resistivity goes DOWN almost linearly with temperature ( causing it to flow even more current and heat more rapidly. (some materials like metals have increasing resistance with temperature). Very quickly it gets to a life threating level of charge in the water. Also, the current dosnt flow in just a straight line between the two ends but uses paths all around (favoring the easiest but sharing).

But why doesn't this level of leakage trip a GFCI?

A GFCI should trip with as little at 5 to 10ma of current detected between the hot and ground. It is likely your Hot to Neutral path in the wetted cord is the "easier" path and taking most of this small current. Did the resistance measurement on the unplugged wetted cord indicate a lower resistance on the Hot to Neutral versus the Hot to Ground (I suspect so)...


Regular tap water will almost never trip the GFCI---salt water will however. I do the extension cord test with students all the time. Regular tap water is just not typically conductive enough. I even have a bucket with a light bulb and socket under water and the light is lit under water and I can reach in the bucket and take the bulb out----all plugged into a GFCI protected receptacle.

  • 1
    If you had your other hand on a ground path, however --such as the sink tap -- this would not be safe without a GFCI (and isn't really "safe" with one, just safer)
    – keshlam
    Sep 9 at 23:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.