Past experiments and modeling with the Tmap code in [1, 2] indicated that BeD co-deposited layers are less (time-wise) efficiently desorbed of retained D in a fixed low-temperature bake, as the layer grows in thickness. In Iter, beryllium rich co-deposited layers will grow in thickness over the life of the machine. Although, compared with the analyses in [1, 2], Iter presents a slightly different bake efficiency problem because of instances of prior tritium recover/control baking. More relevant to Iter, is the thermal release from a new and saturated co-deposit layer in contact with a thickness of previously-baked, less-saturated, co-deposit. Experiments that examine the desorption of saturated co-deposited overlayers in contact with previously baked under-layers are reported and comparison is made to layers of the same combined thickness. Deposition temperatures of ∼ 323 K and ∼ 373 K are explored. It is found that an instance of prior bake leads to a subtle effect on the under-layer. The effect causes the thermal desorption of the new saturated over-layer to deviate from the prediction of the validated Tmap model in [2]. Instead of the D thermal release reflecting the combined thickness and levels of D saturation in the over and under layer, experiment differs in that, i) the desorption is a fractional superposition of desorption from the saturated over-layer, with ii) that of the combined over and under-layer thickness. The result is not easily modeled by Tmap without the incorporation of a thin BeO inter-layer which is confirmed experimentally on baked BeD co-deposits using x-ray micro-analysis.