Hi Ron,
This is very interesting. I only found a little bit of information on KOSREF except that another name maybe KOSOVAREF and that the projected coordinate reference system is the same as EPSG's 3909, MGI 1901 / Balkans 7. The projection parameters are:
Central meridian: 21.0
Latitude of origin: 0.0
Scale factor: 0.9999
False easting: 7500000.0
False northing: 0.0
I assume that the geographic coordinate reference system is MGI AKA MGI 1901, but the offsets between your KOSREF and WGS84 "GK zone 7" coordinates are too small compared to when I convert between MGI and WGS84 using any of the geographic/datum transformations.
I haven't been able to confirm whether KOSREF is on MGI (which uses Bessel 1841 ellipsoid) or not.
Because your offsets are between two projected coordinate reference systems, there isn't a direct way to model the differences. You might be able to do this by creating two projected CRS, and modifying the false easting and false northing parameters of one to incorporate the offsets.
Another method could be to use the Create Custom Geographic Transformation tool. There is a Geographic 2D Offset method. The offsets are in arc-seconds. I unprojected both datasets, assuming that the KOSREF is on MGI 1901.
WGS84 longitude | WGS84 latitude | KOSREF longitude | KOSREF latitude | longitude difference in seconds | latitude difference in seconds |
20.9973292100229 | 42.5403926048346 | 20.9973570639586 | 42.5574121053765 | 0.10027416851415 | 61.2702019508362 |
20.9965745282358 | 42.5377096561408 | 20.9966020849491 | 42.5547288707077 | 0.09920416787566 | 61.2691724408421 |
20.9948579691214 | 42.5377906200380 | 20.9948848524508 | 42.5548098436273 | 0.09677998583015 | 61.2692049214587 |
20.9949186281643 | 42.5404015264134 | 20.9949455363440 | 42.5574210284567 | 0.09686944692788 | 61.2702073558950 |
The offsets aren't as uniform as the projected versions (which seem suspicious to me actually). You could average these and use them.
Melita