Thanks for the detailed post of your experience with the collection data, Mia. I teach cataloging and metadata and one of the most difficult things for students to grasp is why punctuation, vocabulary control and conceptual discreteness is so critical. Your post will certainly help in this regard.
Having spent several years "cleaning" a data set of 30,000 image records at an academic institution I felt your pain! Do you have any plans to speak to someone at the Cooper Hewitt about what you found for verification? I am wondering specifically about the large accessions of textiles in specific years.
Also, I am trying to pull together speakers for a proposal I am working on about visualizing cultural data that will take place next spring. Any interest?