In reworking outlets and switches in our 1960's California home to use AlumiConn connectors + copper pigtails where needed (instead of older wire nuts + Noalox), I came across a duplex wall box in the kitchen that includes incoming and outgoing wires on a circuit that's completely separate from the other wiring in the box. That particular separate circuit only includes outlets, and the wall box is upstream of anything else on the circuit. Disconnecting the wires in that wall box should have left no power whatsoever going to the rest of the circuit ... or so I thought.

After turning off the breaker for the circuit, I started disconnecting the wire nuts connecting this pair of incoming and outgoing wires, and that's when I saw unexpected sparking between the two neutral wires. Nothing was plugged in to the outlets at the time. At first I thought I made a mistake identifying the breakers, but to the best of my abilities, I am certain that the rest of what's in the box is unrelated and separate. I finally proceeded to open up every downstream outlet on that circuit, disconnect all the wires, then test and retest using both a twin-lead tester (kind of like this one) and a continuity tester. All the outlets (superficially at least) appear to be wired in an ordinary manner, with incoming and an outgoing wires and pigtails to the receptacles, much like figure B in this diagram. However, testing reveals that when a certain other breaker in the house (operating lights elsewhere) is turned on, two of the wall boxes on my problem circuit exhibit the following behavior: there is current flowing in one of the neutrals going into the box. My twin-lead tester lights up when I connect it to the neutral and the ground, but not between neutral and hot, and not between hot and ground. Turning off the "other" breaker stops this behavior.

Researching this kind of thing leads me to conclude that the most likely cause is a shared neutral between my problem circuit and another circuit in the house. My best guess is that the wires feeding the two outlet boxes are coming from a junction elsewhere (in other words, that the wiring is not strictly from outlet to outlet as I expected, but that there's unidentified joint somehere), and in that junction, someone connected a neutral from the other circuit to the neutral(s) feeding those outlet boxes. I went into the attic to investigate, but the attic uses loose-fill insulation (so I basically can't see the wiring) and is not easy to navigate safely, so I gave up for now.

Questions: are my conclusions reasonable? What are other possible causes of this behavior?

(More info: there are 2 other things in that wall box, a switch and an outlet, each on separate breakers. The circuit with the outlets is a 3rd one. I am 99.9% certain there are no connections or dependencies between these 3 circuits. Testing was done with all 3 breakers off. The circuit that unexpectedly allows current on the neutrals mentioned above uses a 4th breaker.)

  • First check to see if there are inter-connected neutrals in a box in the living space. It could be that you don't even have junction boxes in the attic. If you do have them, they were supposed to be shielded by a tube from the loose fill insulation or at least have the location marked. Do you have any smart switches in either of the circuits whose neutrals appear to be entangled? Jan 9, 2019 at 19:04
  • @JimStewart Thanks for the suggestion – I'll check the boxes. W.r.t. shielding in the attic, no, unfortunately not in this old house, and even during recent remodeling work, the contractors didn't do that. W.r.t. smart switches: no switches on the circuit at all. There's a GFCI outlet but I made sure to disconnect it completely before the tests described here. On the "other" surprise circuit, there are illuminated (but dumb) switches, which do leak a small current, but as mentioned elsewhere, what I'm puzzled about is seeing a connection between circuits.
    – mhucka
    Jan 9, 2019 at 19:59
  • Did you inspect the attic before the blown in insulation was added? Do you know for a fact that there are junction boxes in the attic? Jan 9, 2019 at 22:29
  • Before you start disconnecting a lot of wires be sure you label them or at least take a picture before you disconnect so that you can restore them to original connections if necessary. There is a simple way to determine if you have hots that are wrongly joined, and a slightly more complicated way to determine if you have wrongly connected neutrals. These methods involve measurements in the panel and you don't have to disconnect anything. You can also do it without removing the dead front of the panel by installing a GFCI receptacle in one of the circuits. If are interested, I will give it. Jan 9, 2019 at 22:46
  • 1
    That's why, here in EU, we use different shades of blue to indicate different neutrals and different colors (but blue and green-yellow) to indicate different live conductors. Just looking at two boxes you instantly see if there's an add-on or a modification.
    – DDS
    Jan 9, 2019 at 23:17

3 Answers 3


First, there is such a thing as a multi-wire branch circuit (MWBC), in which two hots intentionally and consensually shate a neutral. They are placed on opposite poles of service, which means only differential current flows through the neutral. This works for 2 hots in single (split) phase service, and up to 3 hots in 3-phase (208V) service which you will only find commercially or in NYC.

Neutrals must be carefully wired to be monogamous to its partner hot(s), to assure that the neutral carries no more current than its partner hot(s). After all, neutrals don't have breakers. If a neutral carried current for more than its intended hots, it could overload, and nothing would detect this. Borrowing a neutral is not allowed.

Another rule is that currents must be equal in every cable or conduit. This means no loops, wiring routes must be a strict "tree" topology. It also means the neutral will always be in the same cable with its partner hot, or hots in MWBC.

And that means MWBCs will be readily apparent in the service panel, because they will enter the box in a red-black-white cable. (or in the same conduit). When you see that, place both red and black gots on a 2-pole breaker, so that when you switch off one side, the other is also switched off. That satisfies a basic rule of MWBC.

At this point, shutting off that breaker must arrest all current flow in the neutral.

If it does not, then you have a "promiscuous neutral" problem that you should aggressively eliminate.

Grounds, on the other hand, are allowed to be promiscuous, since current only flows on them during fault conditions.


When you turn off a breaker, it opens the hot wire, but not the neutral. Neutrals should not be switched. The sparking you saw between neutrals is not unexpected or a sign something's wrong.

There should never be voltage between neutral and ground. Is it possible that the neutrals where you found voltage to ground but not to hot were actually hots on white wires, such as switch loops?

If there is current flowing on a neutral when it's circuit is turned off, it's likely that that neutral is mixed up with the neutrals of other circuits at some point, which is a problem. The neutrals for all the circuits originating from a panel are bonded at the panel, but they should not be mingled (touching or spliced) anywhere downstream beyond the panel.

  • 2
    But if those are separate circuits, what would be the path from the neutral in one circuit to the neutral in the first circuit, unless the neutrals were joined somewhere?
    – mhucka
    Jan 9, 2019 at 19:40
  • With respect to the possibility of neutrals being hots on a switch loop: unless something is seriously screwed up, I don't think it could be, in this case. I believe I can account for every pair of wires as being wires that go from one outlet to the next. It's not impossible there's something more at work here, but I really tried to narrow down the possibilities. For example, after disconnecting and separating all the outlet wires, I can see that all the lights still work.
    – mhucka
    Jan 9, 2019 at 19:48
  • @mhucka - all the neutrals are bonded at the panel! Jan 9, 2019 at 20:28
  • @batsplatsterson yes, and only at the panel. There is no good reason current would go from one outlet to the panel and through the bond and then back out another circuit's neutral. Maybe in case of a neutral-ground fault, but then the current should be pretty small and not make sparks, I would think? Apr 20, 2023 at 10:06

You have two circuits which you think might be wrongly connected. Don't disconnect any (more) wires and use the AlumiConns to restore the connections you know are correct.

Then test the circuits as follows:

  1. Exclude wrongly connected hots by turning off one breaker and checking to see if hot slots in the receptacles for that circuit are [EDIT] NOT hot. This could be a non-contact voltage tester or just a plug in lamp.

  2. Test for the presence of wrongly connected neutrals in the circuits by installing a GFCI receptacle in one of the circuits in a box as close as you can to the panel. Use the load terminals to feed the rest of the outlets circuit. If you put a load on the end outlet and the GFCI trips, then the neutrals are (wrongly) connected. If it does not trip you probably don't have joined neutrals.

EDIT The way to turn "probably don't have wrongly joined neutrals" into a certainty is to be certain there are no junction boxes (where a wrong connection of neutrals could be made) between where you installed the GFCI receptacle and the panel.

If there could be wrong connections in the attic hidden under the insulation and one of these could be between the first box ([edit] or the box where the GFCI is placed) in the living space and the panel, then you would have to open the dead front of the panel and use a current clamp ammeter to test for current in the neutrals of each circuit when its breaker is off.

There should be no current in the neutral of a circuit with its breaker off. The existence of such current would indicate the neutral is wrongly connected to another circuit which is powered and has active loads so that some (roughly half) of the current is flowing through the neutral path of the circuit being examined.

EDIT If you don't have a current clamp ammeter, you could use a standard ammeter but it would require disconnecting the neutral and inserting the ammeter between the disconnected neutral wire and the neutral bar in the panel.

Before disconnecting the neutral you would switch off both breakers and you would be sure that the total load on the other circuit would not exceed the current capacity of the ammeter when the breaker to the other circuit was switched on to send current down the neutral path.

Expect to use a load which would cause 0.3 A to 5 A total in the neutral path. A single 40 W to 100 W incandescent bulb would do.

EDIT My answer applies if the two circuits are not a MWBC as explained above by @Harper above. If it is a MWBC, then the circuits would share a neutral by design and there would be no reason to change this.

EDITHere is another way to use a GFCI receptacle as a test device to determine whether two circuits have wrongly connected neutrals.

In the panel pick one of the two circuits and switch off its breaker. Disconnect the hot and neutral wires for that circuit and connect them respectively to the hot and neutral load terminals of the GFCI receptacle. Connect pigtails from the breaker and the neutral bar to the respective hot and neutral line terminals of the GFCI receptacle. You have just installed a GFCI receptacle that has no connections between itself and the panel.

Switch the breaker back on and reset the GFCI. Loads on the two circuits will trip the GFCI if there is a neutral connection. If the GFCI does not trip, then there is no neutral connection between the two circuits. (Vary the load on the circuit under test to insure that you haven't by chance gotten balanced loads that are giving a current in the neutral path equal to that in the hot path.)

Your Answer

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

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