Got some collation files back from Cara, and pushed them into the DB. This threw up two problems, not with those files, but with the MS markup file: eventually narrowed those down to one markup error (extra <div> surrounding a poem), and one XSLT problem (<note> tags nested inside the poem were not being processed properly). Also fixed a problem with name lookups; that seems to work OK now on all the poems we've tested.

Checked with Ray re a discrepancy in the requirements for rendering, whereby the old print-style rendering makes effective popups difficult to achieve. The outcome is that we should make the expansive view (where each reading gets its own line) the default, and go for highlighted popups there, while defaulting to an unhighlighted view of the line in the run-together print view.

Cara has found one bug in Apparatus, which seems to stumble over a blank "line" (i.e. an l tag with no readings) in the Collate file. I'll work on that tomorrow.