I have some historic UTM coordinates as NAD27 (556787, 7169250) which I entered into a spreadsheet and have imported into QGIS.

I assign them the projection Nad27 / UTM zone 12N (EPSG: 26712)

I save as... Projection Nad 83 /UTM zone 12N (EPSG: 26912)

When I compare coordinates, the coordinate shift is different than what I know it should be (in particular, there is no east-west component to the transformation), and when I use a separate NTV2 transform there is a shift of ~ 30m north, 50m west.

I have looked at This help article which involves projecting across UTM zones, which I am not doing.

and This other help article which is solved by using a more robust version of the nad27 projection (EPSG: 26712 instead of EPSG:2027) which I am already doing

Am I missing something very simple here?

I am running QGIS 1.8.0 on window

-
No idea where it should be, but see if you can find the NTv2 files. For Canada, you need Ntv2_0.gsb. If it's not there, output is 556784.65 7169444.17. With it, 556732.11 7169475.68. – mkennedy Oct 8 '13 at 15:57
You say "the coordinate shift is different than what I know it should be". How much is this shift exactly and how much you were expecting it? – A.S Oct 8 '13 at 15:57
mkennedy: The new coordinates I obtain are 556784.65, 7169444.17, suggesting I don't have the gsb file. The coordinates you give are what I was expecting. I'm not familiar with gsb files in QGIS. Is there any documentation you can point me to off the top of your head? Ardit: The shift was 194.17m North, 2.34m West. **sorry about the double post. I was not aware of the 5 minute edit limit. – Nick Oct 8 '13 at 16:42
Be aware that the difference between the ellipsoid of NAD27 ellipsoid and the one of NAD83 along the minor axis is 168.5 m and 69.4 along the major axis. en.wikipedia.org/wiki/North_American_Datum – A.S Oct 8 '13 at 17:13
ok, but that doesn't tell you anything about the lat-long or UTM E/N shift... When I use the Ntv2_0 application on the NRCAN website (webapp.geod.nrcan.gc.ca/geod/tools-outils/ntv2.php) I obtain the same results as mkennedy, which is what it should be. I'm just now having trouble getting the transformation tool to work importing the GSB file. – Nick Oct 8 '13 at 17:27