What has a translation error of old papers have to do with the science as we understand it today,
Farsight?
However light does have a velocity!
So
On the Influence of Gravitation on the Propagation of Light
is correct as stated. It is the non-standard usage of "c" as denoting a velocity that may be an issue.
Replacing "velocity of light" with "speed of light" makes the text look better from the modern perspective. No one at the time had a problem with "velocity of light". Only a few crackpots concentrating on semantics have had a problem with "velocity of light" since then. They really go crazy when they realize that
Einstein used "V" for the speed of light until 1907 
!
You are wrong. Einstein used "V" for the speed of light back in 1905. He never uses 'v' to designate 'c'.
I have in front of me an English translation of the article:
'On the Electrodynamics of Moving Bodies' by Albert Einstein,
The German version of it written in 1905. My translation has equations
which I assume were unchanged in the English translation. Just in case
someone wants to challenge me who has the German version, the original
article was
'Zur Elektrodynamik bewegter Körper,' written in 'Annalen de Physik 17 (1905).
Einstein, in this article, designates 'c' for the speed of light in a vacuum. The Maxwell equations that he writes are for a vacuum only, so it makes no sense to discuss the speed of light in glass. He designates 'v' and 'w' for speeds less than the speed of light. He sometimes capitalizes 'V'
when he adds velocities. In section 5, which is page 50 in my edition, he writes the formula for addition of velocities which is:
V = (v+w)/(1+vw/c^2)
I can only address the way he uses English in this translation. However, scientific translators rarely change the variable names in the translation. So if "V" is used to designate a speed less than light in this translation, there is a high probability he used it the same way in the original English article.
I can only address the issue of English words as used by physicists today. I think what Einstein meant was very clear to most physicists in 1905 in the context of the article. However, the physics jargon of the time has since evolved so one can express it clearer.
English speaking physicists usually reserve the word 'velocity' to designate a vector. In conversation they will interchange speed and velocity when the context makes it clear what they are talking about. They use the word 'speed' to designate the magnitude of a 'velocity' vector. In the most rigorous, formal usage of the word 'speed', speed can not be negative. A canonical speed is greater than or equal to 0.
Clearly, 'c' can not be a vector. So calling 'c' the 'velocity of light in a vacuum' is formally wrong. Also, GR shows that the speed of light can vary over large distances in a nonzero gravitational gradient. So in GR, it is only the local speed of light that is invariant.
The variables 'u' and 'v' represent a projection (i.e., a component) of a velocity vector on a designated axis. In the article, the projection is on the 'x-axis'. The quantities 'u' and 'v' can be negative. Therefore, I think that it would be inappropriate to call them speeds. It is probably just as valid to call them velocities. There is a direction simplicity in the values of 'u' and 'v', so they are not really raw 'speeds'. So I propose that Einstein was using the word velocity correctly when he designated 'u' and 'v'.
However, 'c' can not be a vector. So it usually shouldn't be called a velocity. In the context of most equations in the article, 'c' is a speed. It is used this way on page 51, where 'c' designates a vector with the magnitude 'c'.
V = (c+w)/(1+cw/c^2)
so,
V=c
Here, c is any velocity with a magnitude equal to the speed of light in a vacuum. However, c is a vector in this formula! So it is a velocity! The word 'velocity of light' is valid in the context of this expression.
I think this is obvious to educated physicists, who know what a vector is. However, a lot of people criticizing Einstein obviously don't know what a vector is.
There is a simple way to avoid this ambiguity in modern physics classes. I propose the following.
I believe to be 100% consistent with absolutely no ambiguity, 'c' should be designated 'the local speed of light in a vacuum'. The word local excuses the effect of a large gravitational gradient. If you look on a small enough length scale, gravitational gradients are insignificant even in GR!
So the most important postulate in relativity should be stated as follows.
The local speed of light in a vacuum does not vary with reference frame of the observer doing the measurement.
This statement is correct both in GR because it includes the word 'local'. However, it is also valid in SR. So I think that if the postulate is expressed this way, some misunderstandings can be avoided.