Growing evidence supports the notion that plutons are constructed incrementally, commonly over long periods of time, yet field evidence for the multiple injections that seem to be required is commonly sparse or absent. Timescales of up to several million years, among other arguments, indicate that the dominant volume does not remain largely molten, yet if growing plutons are constructed from rapidly solidifying increments it is unlikely that intrusive contacts would escape notice. A model wherein magma increments are emplaced into melt-bearing but crystal-rich host, rather than either solid or crystal-poor material, provides a plausible explanation for this apparent conundrum. A partially solidified intrusion undoubtedly comprises zones with contrasting melt fraction and therefore strength. Depending on whether these zones behave elastically or ductilely in response to dike emplacement, intruding magma may spread to form sheets by either of two mechanisms. If the melt-bearing host is elastic on the relevant timescale, magma spreads rather than continuing to propagate upward, where it encounters a zone of higher rigidity (higher crystal fraction). Similarly, if the dike at first ascends through rigid, melt-poor material and then encounters a zone that is weak enough (poor enough in crystals) to respond ductilely, the ascending material will also spread because the dike tip ceases to propagate as in rigid material. We propose that ascending magma is thus in essence trapped, by either mechanism, within relatively crystal-poor zones. Contacts will commonly be obscure from the start because the contrast between intruding material (crystal-poorer magma) and host (crystal-richer material) is subtle, and they may be obscured even further by subsequent destabilization of the crystal-melt framework. Field evidence and zircon zoning stratigraphy in plutons of the Colorado River region of southern Nevada support the hypothesis that emplacement of magma replenishments into a crystal-laden host is important in pluton construction. The dominant granite unit of the Spirit Mountain batholith displays only subtle internal contacts. However, ages and elemental zoning in zircons demonstrate a protracted history of almost 2 million years, major fluctuations in T and host melt chemistry, and mixing of strongly contrasting populations of magmatic zircon in single samples. We interpret this to reflect reactivation of rigid sponge and mush and entrainment of earlier-formed crystals, and we infer that this was in response to granitic replenishment. Much of the smaller Aztec Wash pluton comprises interlayered cumulate-textured quartz monzonite and mafic sheets. The latest phase of pluton emplacement is marked by numerous thick, fine-grained granite "sills" that intruded the subhorizontal quartz monzonite sheets. Contacts between granite and quartz monzonite are "soft," highly irregular on cm-dm scale with coarse xenocrysts from the quartz monzonite entrained in the fine-grained granite. We interpret the granite replenishments to have spread laterally within crystal-rich, melt-bearing quartz monzonite beneath rigid mafic sheets. In this case, clear evidence for the emplacement process is fortuitously preserved because the granite was emplaced in the waning stage of the thermal lifetime of the pluton, and because the mafic sheets enhance the strength contrast and make the geometry more visible. Similar "sills" of fine-grained granite were also preserved during the late stages of the history of the Spirit Mountain batholith. ?? 2009 Elsevier B.V.