Yep, that's true, but only in the zero sequence is there enough coupling to matter. Looking at one section of a two line right-of-way picked at random from our line data I see the zero sequence mutual coupling being 0.00018+0.00062j while the other mutuals (nine total for two lines) range from 8.36777*10^-7+4.80292j*10^-6 to 5.26882*10^-6+4.56966j*10^-6. With the uncertainties in modeling, the two or three orders of magnitude difference means that the rest are almost always ignored.