If you sized a buried cable last year, you used Reference Method D. One column in the current carrying capacity tables covered both a cable pulled through underground ducting and a cable laid directly in the ground with sand and marker tape over it.
Those are not the same thermal situation, and Amendment 4 finally says so. BS 7671:2018+A4:2026, published April 2026, splits Method D into two methods across Tables 4D4A, 4E4A, 4H4A, and 4J4A:
D1: in ducting in the ground. Installation method 70.
D2: direct in the ground. Installation methods 72 and 73.
If your design basis, your spreadsheet, or your software still offers a single "Method D," it is now specifying an installation method that the standard no longer contains.
Why the split exists
A buried cable sheds heat into the surrounding soil. Soil, particularly when moist, conducts that heat away well. A cable inside a duct sheds heat first into an air gap, then through the duct wall, then into the soil. The air gap is the problem. Air is a poor thermal conductor, so the ducted cable runs hotter for the same current.
The old single Method D column had to cover the worst case of the two, which meant a direct buried cable was rated as though it had a duct's thermal penalty it does not actually have. The split removes that compromise. D2, direct buried, now carries the higher ratings the installation genuinely supports. D1 carries the ducted figures.
What it means in numbers
Take a common sub main: 70 mm² four core SWA, copper, 90 degree thermoset, three phase.
Under the pre Amendment tables, Method D gave a tabulated capacity of 167 A regardless of whether the cable was ducted or direct buried.
Under A4, that installation now resolves to one of two figures:
- D1, in ducting: 167 A (installation method 70)
- D2, direct buried: 188 A (installation methods 72 and 73)
The practical consequence runs in one direction. Direct buried designs sized under the old combined column were carrying a hidden margin, and re running them under D2 can release capacity, sometimes enough to drop a cable size on a long run. Ducted designs get no such gift, and a designer who habitually treated "buried" as one condition needs to confirm which side of the split every existing detail actually sits on.
The corrections that still apply on top
The split changes which column you start from. It does not change the stack of corrections that buried installations attract, and these are the ones I see missed most often in checked designs:
Soil thermal resistivity, Table 4B3. The tabulated figures assume 2.5 K.m/W. The correction is not symmetrical between the two new methods, which is itself evidence for why the split matters. A direct buried cable in favourable wet ground at 1.0 K.m/W earns a rating factor of 1.5, while the same soil lifts a ducted cable by only 1.18. At the dry end, 3.0 K.m/W pulls direct buried down to 0.9 against 0.96 for ducts. If nobody on the project measured or declared a soil resistivity, the design is carrying an assumption someone should own.
Depth of laying, Table 4B4. Ratings reference 0.7 m. Deeper burial derates modestly but measurably, down to 0.92 for a direct buried cable at 2 m, while a shallow 0.5 m run actually gains a few percent.
The 0.9 factor for overload protected buried circuits. Where a buried cable relies on its protective device for overload protection, a further 0.9 rating factor applies. This one is routinely forgotten because it lives in the Appendix 4 preliminary text rather than in a table.
A direct buried cable can therefore gain from the D2 column and lose it all back to poor soil, depth, and the 0.9 factor. The split makes the starting point honest. The corrections still decide the answer.
A detail from implementing this
While building the A4 tables into CurrentProof, one anomaly stopped us. In Table 4D4A, at 1.5 mm² and 2.5 mm², the D2 figures largely fail to exceed D1, with one cell actually falling below it, which breaks the pattern the physics suggests and the rest of the tables follow. We held those two cells out of the database, re checked them against the published source twice on separate occasions, got identical figures both times, and only then populated them. The anomaly is real, not a transcription error.
I mention it for one reason. Every designer moving to A4 is about to re key numbers, update spreadsheets, or trust a software vendor who did. The two cells that look wrong are exactly the ones worth tracing to the printed page.
The deadline
A3 is withdrawn on 15 October 2026. From that point, new designs need A4 as their basis, and "we have not updated the tables yet" stops being an inconvenience and becomes a compliance gap. Through the changeover, the D1 and D2 split is the change most likely to be silently missed, because a design that still says "Method D" looks complete right up until someone asks which method 70, 72, or 73 it meant.