Yes, Mike has already corrected his code,
So when you apply that fix and you arrive at the line "* This is one of the lines that you want to see"
Then you are in the scan loop at the record from which the memo is, so you have access to Part# and all the data from the memo line is in laLines(lnI). You can split this sting line again into single values, either using GETWORDNUM(), STREXTRACT() or again use ALINES with "," as the "line"-seperator. So ALINES cannot only split a multiline memo/text/string into lines, you can split at any seperator you specify as parameter. The single elements you get are also strings, but you can easily translate "T" and "F" to .T. and .F. of course, EVAL("."+Flagvalue+".") would be one way to do that, when Flagvalue is the "T" or "F" you extract from the line string.
So overall, you are in the process of normalising data already, then. I would perhaps just simply cut this into the esier steps to first simply add any line to a detail table and afterwards remove unwanted detail records by a simple DELETE-SQL query, if that makes any sense at all. It's of course easy enough to skip all rows not cointaining "T", but it would of course be fatal, if your first string column (in your example line the "1234" would also contain a T. Therefore I'd prefer to filter after the data is translated into the single field values you need.