bah.txt

/home/jbfink/.rbenv/versions/2.1.2/gemsets/ruby-marc/gems/marc-0.8.1/lib/marc/marc8/to_unicode.rb:163:in `rescue in transcode': MARC8, input byte offset 20, code set: 0x45, code point: 0xa0 (Encoding::InvalidByteSequenceError)
	from /home/jbfink/.rbenv/versions/2.1.2/gemsets/ruby-marc/gems/marc-0.8.1/lib/marc/marc8/to_unicode.rb:144:in `transcode'
	from /home/jbfink/.rbenv/versions/2.1.2/gemsets/ruby-marc/gems/marc-0.8.1/lib/marc/reader.rb:397:in `set_encoding'
	from /home/jbfink/.rbenv/versions/2.1.2/gemsets/ruby-marc/gems/marc-0.8.1/lib/marc/reader.rb:359:in `block (2 levels) in decode'
	from /home/jbfink/.rbenv/versions/2.1.2/gemsets/ruby-marc/gems/marc-0.8.1/lib/marc/reader.rb:358:in `each'
	from /home/jbfink/.rbenv/versions/2.1.2/gemsets/ruby-marc/gems/marc-0.8.1/lib/marc/reader.rb:358:in `block in decode'
	from /home/jbfink/.rbenv/versions/2.1.2/gemsets/ruby-marc/gems/marc-0.8.1/lib/marc/reader.rb:307:in `upto'
	from /home/jbfink/.rbenv/versions/2.1.2/gemsets/ruby-marc/gems/marc-0.8.1/lib/marc/reader.rb:307:in `decode'
	from /home/jbfink/.rbenv/versions/2.1.2/gemsets/ruby-marc/gems/marc-0.8.1/lib/marc/reader.rb:247:in `each'
	from slice.rb:3:in `<main>'

It's the Harvard dataset here: http://openmetadata.lib.harvard.edu/bibdata . file / magic sez it's MARC-21 Bibliographic.

Binary totally works and is probably close enough for government work. I don't want (at this stage) to mess with Catmandu or other things to convert it -- maybe later. Thanks man, you're the best.