Context : One of the greatest uncertainties in modelling the mass-exchange phases during the evolution of a binary system is the amount of mass and angular momentum that has been lost from the system .
In order to constrain this problem , a favourable , evolved and detached real binary system is valuable as an example of the end result of this process .

Aims : We study the 52-day post-mass-exchange eclipsing binary V643 Ori from complete uvby light curves and high-resolution spectra .
V643 Ori is double-lined and shows total primary eclipses .
The orbit is accurately circular and the rotation of both stars synchronised with the orbit , but the photometry from a single year ( 1993 ) shows signs of weak spot activity ( 0.02 mag ) around the primary eclipse .

Methods : We determine accurate masses of 3.3 and 1.9 M _ { \sun } from the spectroscopic orbit and solve the four light curves to determine radii of 16 and 21 R _ { \sun } , using the Wilson-Devinney photometric code .
The rotational velocities from the cross-correlation profiles agree well with those computed from the known radii and orbital parameters .
All observable parameters are thus very precisely determined , but the masses and radii of V643 Ori are incompatible with undisturbed post-main-sequence evolution .

Results : We have attempted to simulate the past evolutionary history of V643 Ori under both conservative and non-conservative Case B mass transfer scenarios .
In the non-conservative case we varied the amounts of mass and angular momentum loss needed to arrive at the present masses in a circular 52-day orbit , keeping the two stars detached and synchronized as now observed , but without following the evolution of other stellar properties in any detail .
Multiple possible solutions were found .
Further attempts were made using both the BSE formalism and the binary MESA code in order to track stellar evolution more closely , and make use of the measured radii and temperatures as important additional constraints .
Those efforts did not yield satisfactory solutions , possibly due to limitations in handling mass transfer in evolved stars such as these .
We remain hopeful that future theoreticians can more fully model the system under realistic conditions .

Conclusions :