PDF Copy Doubles Letters? Fake Bold Is the Cause

When a PDF copy doubles letters, so that "Total" pastes as "TToottaall", the text was almost always drawn twice to fake a bold weight. The file has no bold font, so it prints each word two times with a tiny offset. Copying picks up both copies, interleaved. A per-word formula in Excel halves them back.

Why does a PDF copy double letters like TToottaall?

A PDF copy doubles letters because the page holds two copies of the same text, placed almost exactly on top of each other. Drawing a word twice, shifted by a fraction of a point, makes it look heavier. Some PDF producers use this to imitate bold type when no bold font is embedded.

On screen the trick works. The two copies overlap so closely that you see one slightly thicker word. A copy tool sees something else: two characters at nearly the same position for every letter. It sorts characters by where they sit on the line, and since each pair shares a spot, the pairs come out side by side. "Total" becomes T, T, o, o, t, t, a, a, l, l.

That is why the doubling has such a regular shape. Every character appears twice in a row, including digits and punctuation. Spaces may or may not be doubled, depending on whether the producer drew them. Occasionally a producer draws the text three or four times for an even heavier look, and you get tripled letters by the same logic.

Spotting fake bold text before you copy the PDF

Fake bold usually affects only the lines meant to look bold: headings, column titles, totals and sometimes account names. Body text in a regular weight copies normally. So before copying a whole page, test one bold line and one ordinary line and compare what you get.

Three quick tests:

• Paste a heading into a plain text editor. If it reads "AAccccoouunntt ssuummmmaarryy", the line was drawn twice.
• Zoom in hard on a bold word. At high magnification, fake bold often shows a faint second outline or edges that look smeared sideways.
• Use the reader's search. Search for a bold word such as "Total". A search that finds the regular-weight copies but misses the bold ones hints that the bold text is stored differently.

Doubling is one of several ways copied PDF text goes wrong, and they can stack. Words that run together with no gaps have a different cause, explained in why PDF copy gives no spaces between words. Lines that come out in a strange sequence are a reading order problem, covered in why PDF reading order comes out scrambled.

Cleaning doubled PDF letters with an Excel formula

Clean doubled letters with a formula that checks each word and halves it only when the word is exactly doubled. Halving every cell blindly would damage normal text, because plenty of real words contain repeated letters. Checking word by word keeps "Coffee" intact while fixing "CCooffffeeee".

In Excel 365, with the pasted text in A2:

=TEXTJOIN(" ", TRUE, MAP(TEXTSPLIT(A2, " ", , TRUE), LAMBDA(w,
  LET(n, LEN(w), h, INT(n/2),
    odd, CONCAT(MID(w, SEQUENCE(MAX(1,h),,1,2), 1)),
    even, CONCAT(MID(w, SEQUENCE(MAX(1,h),,2,2), 1)),
    IF(AND(n > 1, ISEVEN(n), EXACT(odd, even)), odd, w)))))

The formula splits the cell into words, ignoring empty pieces left by doubled spaces. For each word it builds two strings: the characters in odd positions and those in even positions. If the word has an even length and both strings match exactly, the word was doubled, and the odd-position string is kept. Otherwise the word is returned unchanged.

A worked example

The table shows doubled text from an example statement and what the formula returns. The values are made up for illustration.

One edge case slips through. A real word made of one letter pair, such as "AA" in an airline code, also passes the test and would become "A". Such words are rare in statement text, but scan the results for very short words. If your file triples letters instead, change the test to compare three strings, or ask a converter to read the file instead.

Doubled PDF digits in amounts and dates

Doubled digits are more dangerous than doubled letters because a doubled number can look like a different, valid number. "TToottaall" is plainly broken. "1100" could be a doubled "10" or a genuine one thousand one hundred. Only the page tells you which, and a formula cannot.

Treat amounts and dates as a separate check:

1. Clean only the rows you know are doubled. Usually these are total and subtotal lines in bold, not ordinary transactions.
2. Compare cleaned totals with the page. A doubled total halved correctly should match what the PDF shows.
3. Run the balance arithmetic. Opening balance plus credits minus debits should equal the closing balance. A wrongly halved amount breaks it at once.

If the doubled text landed in merged cells after a paste or conversion, unmerge first so each value sits in its own cell before you run any formula. Unmerging cells after PDF conversion shows how to do that and fill the gaps.

Will a converter clear the doubled PDF letters?

A converter can avoid doubled letters that copy and paste cannot, because it sees the position of every character, not just their sequence. Two identical characters drawn at almost the same spot are a strong sign of fake bold, and some extraction tools drop the duplicate before building words.

Not every tool does this, so test a page with bold headings or totals before trusting a converter with a whole file. Look at the bold rows in the result first. If they are clean, the tool handled the duplicates; if they are doubled, clean them with the per-word TEXTSPLIT formula.

FastBankConvert is built to read statement layouts into date, description, amount and balance columns, and it warns you before download when the balances do not add up, which is how a doubled amount would show. It is getting ready to open; join through Get started to hear when trial conversions are available.