I'm working on a friend's house to replace a ceiling light fixture controlled by a wall switch. In the process of trying to figure out which breaker de-energizes the fixture, I've determined that TWO of them do: in a logical-or manner, that is, if either or both breakers are "on", then the fixture is energized (I must turn off BOTH breakers to de-energize the fixture). The two breakers are both single-pole and are not neighbors; they are 6 spaces apart on the same side of the panel (same column of breakers).

I tore into the fixture and there are three 12-2 NM cables coming into its junction box. Disconnecting them all and testing: #1 is energized by one breaker, #2 is energized by the other breaker, and #3 is a switch leg/loop (zero ohms when the switch is "on" infinite when it's "off"). The hots from #1 and #2 were both tied to #3, and the other conductor in #3 to the light's hot input. I can only conjecture that the original electrician thought #2 was just a continuation of the circuit of #1. Oddly too, there seem to be no other loads (with both breakers off, I can't find anything else in the entire house that is de-energized).

Is there any valid reason for this to be wired this way ? What should I do to rectify this inappropriate wiring ? Simply remove #1 or #2 from the fixture and tie it off in its own box ?

FWIW, I've found a couple of other profoundly stupid things in this house: an extension cord that was repaired with the 3 conductors scrambled (not just hot and neutral reversed, which is bad enough), and a luminaire where the wires are just poked into the attic and taped to the romex (not inside a box, and using masking tape). But ... this is the first screwup that looks like it was done by the original electrician.

  • 1
    I wonder if the "second circuit" cable was actually a continuation of the first circuit, and it goes onward to another switch box where the second circuit is present, and they're bridged there because the electrician just tied all neutrals and all always-hots. That is to say, the improper joining of the two circuits might not have happened here: it could have happened anywhere along this "ring circuit" to borrow the British term. Commented Jan 21, 2022 at 4:42
  • Good point. But if in fact it IS a ring, it doesn't matter where I break it (as far as safety or of practicality) - and for that matter, the question of where the "improper joining" took place is really a matter of semantics (for lack of a better term). Unless of course putting the discontinuity in the wrong place overloads one of the circuits. But since I can't identify a single other load on this circuit (or circuits) other than the light in question, that is doubtful. Commented Jan 21, 2022 at 5:32
  • Anyhow, the house hasn't burned down in the 35 years it's stood. But I just don't want to make things WORSE by re-wiring this light. Oh, and did I mention it's a Federal Pacific panel (no joke) ... Commented Jan 21, 2022 at 5:34
  • I re-titled and re-worded my OP to focus on the remedy. Commented Jan 21, 2022 at 5:42
  • OK I edited my answer. TLDR: map the loop and figure out the best place to break it - likely someplace where 2 circuits would be really useful... that being why the original guy ran 2 circuits there. Commented Jan 21, 2022 at 9:43

3 Answers 3


Edit: What you have is an "accidental ring circuit". Somewhere, in some box (clearly not this one), both circuits arrived into the same junction box. That's fine as long as a Great Wall is kept between the two circuits' hots and neutrals. (grounds should be combined). However, at one box or another, someone merged the circuits by accident.

My general view is "Previous work was done that way for a really good reason, even if it got broke later". So in an "accidental loop" situation, I would map the entire circuit(s) to find all the boxes that are served by this "loop". In one of those boxes, it will be obvious and probably, even advantageous to have that box served by 2 circuits instead of one (i.e. the reason they did that). I would break the loop there. For the most advantage.

That's very likely what has happened. Somebody brought a "switch loop" into a box containing an unrelated circuit or branch of a circuit.

Code requires you keep a "Great Wall" between the unrelated "circuits", they don't go as far as requiring a physical divider, but maybe they should.

And then fools get in there and start muxing around, and stuff like this shows up. It's the old "try things until you find something that works" fallacy that I often warn people about; lots of arrangements will work but then will set the stage for problems later.

Not the least of the problems here is that you have 40A worth of hot power that could potentially be trying to return on a single 20A neutral wire. (or 30/15 if applicable). There are also issues with magnetic fields from current not being balanced/equal in cables, all sorts of "wrong".

I agree - a careful examination of the wiring is called for. I recommend a viewpoint of "the people doing this weren't complete idiots, they surely were trying to go for something here".

One fast way of validating a given circuit is to temporarily put its hot and neutral wires on a GFCI breaker. If there's any improper crossing of circuits going on, the GFCI will trip the instant a load is applied. One must be careful of MWBCs, but an MWBC can be validated (as isolated from other circuits) by putting both its hots on the GFCI hot lug.

  • 1
    That's a cool idea about temporarily using a GFCI Commented Jan 19, 2022 at 8:22
  • If it were up to me, I'd outlaw MWBC. Just to save 25% on copper costs ? Commented Jan 19, 2022 at 8:23
  • 1
    This would be better if it didn't begin with "Yes" (and repeats "Yes" in another paragraph) when the titular question is one you're really answering with "No". While you do explain you're agreeing with the question author's analysis of what probably happened to get to this point, the current wording could be confusing to someone not reading closely. It doesn't help that you never directly answer the titular question. I think, perhaps, the issue is you wrote it from, and a lot of us read it from, the point of view of "this is obviously and clearly wrong", so you just jump into an explanation.
    – Makyen
    Commented Jan 19, 2022 at 15:12
  • 1
    @JosephSible-ReinstateMonica A worthy question. But when you merge a MWBC so both hot legs are fed from the same breaker lug, effectively it stops being a MWBC, and current is limited by that breaker. So the sum of the two #12 hots will be say 20A, and the neutral is limited to 20A. Commented Jan 19, 2022 at 18:19
  • 1
    @RustyShackleford many of those manufacturers sell the appliance in the Philippines, which is wired like us, but without neutral. They install a simple transformer to serve the internal 120V loads. But oven light is 220V so they can use indigenous bulbs. Commented Jan 20, 2022 at 0:33

The live "tied together" from two breakers is definitely not allowed, but two different circuits in one outlet box is allowed. Disconnect the tie.

  • I'm not sure "tied together" is the situation, but I can't think of another explanation. But I'm going to tear into the thing soon. Commented Jan 19, 2022 at 6:21

I have a similar thing in my house. Two separate breakers control one light. The light in question is on a 3 way switch. I believe they share a neutral. I discovered the issue when trying to install a GFCI outlet (which wasn't possible, because it would always trip). I think the issue was cheap/lazy wiring standards when the house was built/remodeled. Its a problem, but not a problem that needs immediate fixing. (unless you already have stuff torn apart)

  • 3
    Better to use the GFCI problem as an excuse to hunt down the root problem. Likely the original wiring can be salvaged by use of smart switches. Or, you can still install the GFCI receptacle, you just have to contain the urge to use the "LOAD" terminals for the extra wires. Read the instructions (required anyway by NEC 110.2) and it will say how to hook 2 wires to Line screws. Commented Jan 19, 2022 at 18:31

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.