I've seen photos of "color Daguerreotype" images (they weren't called Heliochrome, at least in that article) in a 1990s vintage Scientific American I have around here somewhere -- it was a history of early color photographic processes. The process, as set forth there, was to put a silver chloride emulsion directly onto glass and expose through the glass with the emulsion in contact with liquid mercury. There was little more said about it, but the images presented were very colorful.
It looked to me as if it would be simpler to put the silver chloride onto a silver plated layer on the glass, expose from the emulsion side, and thus keep the reflective surface with the image. However, where the description stumbled, for me, was in lacking an explanation of how interference layers (essentially in-situ dichroic filters) could form without some kind of support, or how silver (even developed as a Dag) could become reflective enough to produce a dichroic effect when forming at a molecular scale from reduced halide. Given that hand coloring Dags was quite common (though the colors used were traditionally quite pale) but Dags were never transparent, I'm led to wonder if the Heliochrome plates weren't in fact early transparent negatives (i.e. collodion wet plate), colored on the collodion surface; viewing by relfection would intensify the colors and make the areas that would be light in a tintype or ambrotype dark.
I suppose its even possible that a collodion emulsion exposed the way the Heliochrome was said to be made could support interference colors when viewed by reflection; there would need to be enough emulsion thickness to support multiple-layer interference, but I think the sensitization of collodion penetrates more than deeply enough for this, and the grain in collodion negatives is plenty fine...
