The shelf seas of Antarctica are controlled by a combined sea ice-ice shelf system.  Determination of sea ice coverage is crucial for reliable planetary-scale simulations of the ocean-atmosphere system.  Estimation of ice shelf longevity is vital for sea level predictions.  However, they are a connected system.  Despite rising global temperatures and clearly declining sea ice in the Arctic, Antarctic sea ice coverage is increasing.  The ice shelves that comprise 40% of the Antarctic coastal margin represent one driver of this hemispherical difference.  Outflow melt-driven plumes from the oceanic cavities beneath these shelves inject supercooled seawater into coastal surface waters.  This supercooled water, a result of melting of the shelf underside at depth, drives sea ice growth by (i) trapping boundary transfer of heat through increased stratification as well as (ii) removing heat from the ice into the ocean.  It is now recognised that ice crystallisation within the water column is a crucial part of this sea ice development, with columnar ice augmented by platelet ice crystals that grow in the supercooled water.  Here we seek to challenge the sea ice growth paradigm using observations of pelagic mature crystals in an outflow region from the Ross/McMurdo Ice Shelves.  A decade of observations has provided a connecting patchwork of evidence that the ice crystals can exit the shelf in suspended form.  We targeted this issue with a fast ice field camp that captured the oceanic temperature structure, the suspended crystals and the aggregated depositional layer.  These suspended crystals are far larger than expected and as they are depositing substantially, or even fully, grown, it affects the nature and distribution of the crystals on the sea ice underside.  This influences interfacial momentum transfer, sea ice composition and ecological habitat throughout large parts of the Antarctic coastal margin.