Key points are not available for this paper at this time.
The MOND paradigm of modified dynamics predicts that the asymptotic gravitational potential of an isolated, bounded (baryonic) mass, M, is (r) = (MGa₀) ^1/2ln (r). Relativistic MOND theories predict that the lensing effects of M are dictated by (r) as general-relativity lensing is dictated by the Newtonian potential. Thus MOND predicts that the asymptotic Newtonian potential deduced from galaxy-galaxy gravitational lensing will have (1) a logarithmic r dependence, and (2) a normalization (parametrized standardly as 2^2) that depends only on M: = (MGa₀/4) ^1/4. I compare these predictions with recent results of galaxy-galaxy lensing, and find agreement on all counts. For the ``blue''-lenses subsample (``spiral'' galaxies) MOND reproduces the observations well with an r^'-band M/Lₑ^{'} (1--3) (M/L) _, and for ``red'' lenses (``elliptical'' galaxies) with M/Lₑ^{'} (3--6) (M/L) _, both consistent with baryons only. In contradistinction, Newtonian analysis requires, typically, M/Lₑ^{'}130 (M/L) _, bespeaking a mass discrepancy of a factor 40. Compared with the staple, rotation-curve tests, MOND is here tested in a wider population of galaxies, through a different phenomenon, using relativistic test objects, and is probed to several-times-lower accelerations--as low as a few percent of a₀.
Mordehai Milgrom (2013) studied this question.