To facilitate a comparative study of approaches and algorithms for two-dimensional interpretation of magnetotelluric (MT) data, a dataset of thirty-five sites, called COPROD2, was distributed to the electromagnetic induction community. The data are from stations along a 400 km east-west profile in southern Saskatchewan and Manitoba, Canada, crossing the thick Paleozoic sediments of the Williston basin. Within the basement beneath the sediments lies one of the world's longest and most enigmatic crustal conductivity features -- the North American Central Plains (NACP) conductivity anomaly. Also, at the eastern extremity of the profile is a second basement anomaly (TOBE) interpreted to be associated with the Thompson Nickel Belt at the Superior-Churchill boundary. The MT data were corrected for static shifts, and only the off-diagonal impedance data together with the transfer function data, and their errors, were made available. These MT data are of wide bandwidth (384 Hz -- 1820 s) and high quality (impedance errors typically <2%), and require sophisticated modelling and inversion in order to extract as much meaningful information as possible from them. A challenge for those interpreting these data is that there is a very small, but critical, response in the B-polarization data to the presence of the basement anomalies. In this introductory paper, the previous electromagnetic, and other geophysical, studies of the NACP anomaly, and its tectonic setting within the Trans-Hudson orogen, are described. Representative data from the COPROD2 data are illustrated, and models derived by various groups are shown. Comparisons of these models reveals that a single measure of misfit is an inadequate description of how well a model fits a dataset; one must compare the model fit to the data at virtually each datum to avoid trends in the misfit residuals. Additionally, reliable and consistent error and static shift estimates are essential in order to obtain high resolution images of the Earth's conductivity structure. Finally, the COPROD2 data highlight the sensitivity in B-polarization mode data to breaks in the resistive layer, whereas the E-polarization data sense regions of enhanced conductivity.