Thank you Matt for the thoughtful update.
As I suspected, after the 400 or 450 hours mark, Aeris would start a gradual normalization of treble artifacts. I still expect some fluctuations throughout the audible bandwidth for a while. My experience with Aeris, as well as other Rowland components is that they are late bloomers... They continue to break-in and bloom at least until the 1200 hours mark. At 700 to 800 hours, they give a semi-reasonable idea of their capabilities, but they still do not yield their best.
BTW, I have confirmation from the Rowland factory that Aeris delivers its best performance through its SPDIF inputs, which are transformer-coupled for optimizing common mode noise rejection, galvanic isolation, and impedance stability... This will be particularly effective if Eris is fed from a transformer-coupled source. It is interesting to note that Rowland feeds Aeris mostly through the Bryston BDP-2 server/player, which has implemented transformer coupling on its SPDIF and XLR outputs.
No, I am not using the Bryston player in my system yet... I am a dinosaur, and am still using the Esoteric X-01 as a source.
G.