NAD 27 and NAD 83 are the two horizontal datums that most existing American survey records are tied to. Between them, essentially everything changed: the ellipsoid, the way the datum was positioned relative to the Earth, the measurements it was built from, and the mathematics of the adjustment. The visible consequence is that the same physical monument has two different latitudes and longitudes, differing by tens of meters and by an amount that varies across the country.
Because the shift is not a constant, you cannot correct an old plat with a single offset. That single fact is the practical content of this page.
NAD 27
The North American Datum of 1927 was built on the Clarke 1866 ellipsoid, which has a semi-major axis of about 6 378 206.4 m — roughly 69 m smaller than GRS 80 — and a reciprocal flattening near 294.98. The datum was oriented by holding a single station, Meades Ranch in Kansas, fixed in position and azimuth. Everything else in the network was computed outward from there through triangulation.
That approach makes an ellipsoid that fits North America well and the rest of the planet poorly, which was the correct engineering trade in 1927. The origin is not at the center of mass of the Earth and was never intended to be. The adjustment itself was carried out by hand and by mechanical calculation over decades, in sections, with the practical result that errors accumulated and were absorbed unevenly across the continent — the network is internally strained.
NAD 83
The North American Datum of 1983 was built on GRS 80, with a semi-major axis of 6 378 137 m and a reciprocal flattening of 298.257222101, and was intended to be geocentric. It came from a simultaneous least-squares adjustment of hundreds of thousands of stations, incorporating not just classical triangulation but Doppler satellite positioning and very long baseline interferometry. It was a genuinely different kind of computation, not a refinement of the old one.
The result is a far more internally consistent network. It is also a network whose coordinates simply do not match the old ones anywhere.
What actually changed
| Property | NAD 27 | NAD 83 (original 1986 adjustment) |
|---|---|---|
| Ellipsoid | Clarke 1866 | GRS 80 |
| Semi-major axis | ≈ 6 378 206.4 m | 6 378 137 m |
| Reciprocal flattening | ≈ 294.9787 | 298.257222101 |
| Origin | Held at Meades Ranch, Kansas | Intended geocentric |
| Primary observations | Triangulation, traverse, astronomic azimuths | Classical plus Doppler and VLBI |
| Adjustment | Sectional, over decades, by hand | Single simultaneous least-squares adjustment |
| Best fit | North America only | Global, with a North American emphasis |
How far did coordinates move?
There is no single number, and that is the point. Across the conterminous United States the horizontal shift between NAD 27 and NAD 83 typically runs from a few tens of meters up to around 100 m, and both the size and the direction change with location. In parts of Alaska and the Pacific territories the shift is much larger still.
The shift has two components superimposed. One is systematic: a different ellipsoid and a different origin move everything in a smooth, predictable pattern. The other is not systematic at all: it is the accumulated distortion in the NAD 27 network, which varies irregularly over distances of tens of kilometers. The systematic part could be handled with a seven-parameter transformation. The irregular part cannot, which is why the standard tool is a gridded model.
State Plane made the change more visible
The State Plane Coordinate System was redefined at the same time. SPCS 27 was defined in feet with zone constants computed on Clarke 1866; SPCS 83 was defined in meters on GRS 80, and several states changed the projection or the zone layout in the process. Michigan is the standard illustration: transverse Mercator zones under NAD 27, Lambert conformal conic zones under NAD 83.
The result is that a State Plane coordinate pair can differ between the two systems by hundreds of thousands of units, because the false origins changed as well. That is a good thing — the numbers are so obviously different that nobody confuses them. Latitude and longitude are the dangerous case, because a NAD 27 latitude and a NAD 83 latitude look equally plausible and differ by an amount that will lose a corner.
Working with legacy records
- Establish which datum the record is on before doing anything else. Look for the year, the zone name, the units, and whether values are in feet or meters.
- Treat an unlabelled coordinate on an old plat as unknown, not as NAD 27 by default.
- Prefer recovered physical monuments to transformed coordinates. The monument is the boundary; the coordinate was always only a description of it.
- When you must transform, use NADCON rather than a fitted constant shift, and record which version you used.
- Report the transformed values as transformed, with the source datum, target datum and method noted.
It is worth remembering why this matters beyond arithmetic. A retracement follows the footsteps of the original surveyor, and those footsteps are monuments and calls, not coordinates. NAD 27 coordinates on an old plat are evidence of where the original surveyor thought the corner was, in a frame that has since been shown to be internally strained. They inform the search. They do not settle it.