Kevin Furst: unfortunatley, I see you discussion just now, but maybe it is helfpul for other(s)/other issues. I usually find to spot such typoes easier in marcxml. first of all a space is easier visible in xml due to the `<...></...>`. (if it's just one line of xml `xmllint --format` comes in handy to get it readable for humans.) furthermore, one can the use selects combined with `sort|uniq` on the shell.
eg. in your case running something like
```
$ xmlstarlet sel -t -v '//record/datafield[@tag="952"]/subfield[@code="b"]' /tmp/4koha.xml | sort | uniq
```
would give a list of all _unique_ entries of `952 |b`.