Thank you for spotting this. The transcriptions for the Newton papers are supplied by a third party (The Newton Project, http://www.newtonproject.sussex.ac.uk) so we will pass this onto them.
At a quick glance at a few pages it seems browsers are rendering various sections differently which may account for some errors.
Thanks again,
Cambridge Digital Library
The page reads (from the manuscript image, interpreting the abbreviations) "[...] an arithmetic progression increasing from an unite by 1 composeth triangles by 2, composes squares by 3, composes pentangles by 4, hexangles &c. as 1.2.3.4.5.6. composes the triangles [...]".
In the "Transcription (normalised)" this is
"[...] an arithmet: progres: increasing from an unite by b=2 formula composeth triangles. by a=5/3, composes squares. by y=22/61, composes pentangles. by x=33/61, hexang: &c as 1 compose the triangles 2 &c likewise 3 compose 4 &c So 1.2.3.4.5.6. compose the quintangles [...].
It appears nearly all the 'MathML formulas' are wrong? There is also a textual transcription error "quintangles" which should read "triangles".
This is the only page I looked at. The "Transcription (diplomatic)" appears to bear the same errors. If this page is typical I hate to think how the hard to read or complex mathematical pages have been rendered in transcription.
Edit: I've just noticed that the erroneous MathML formulas are correct renderings of other expressions on the same page, this is probably a coding/markup error?