Uh, no smartass it's merely the retelling of an anecdote not a carefully controlled scientific experiment designed to eliminate every variable. The agent, as the story was told to me, was standing in front of the passenger side of the vehicle, body brushing up against the vehicle, holding the gun in his right hand, shooting at a point roughly consistent with the passenger position, thus, an angle of close to 90% in the horizontal plane. However most windshields, even in the 1960s, have a backward rake, which varies from model to model, but which means that in the vertical plane, there will be a signficant but unknown angle. Now I know my own father, and he's not a liar. If he says it happened, then it happened, as far as I am concerned. I don't know what specific variable combined to make it happen. But I do have independent data showing how unexpectedly low the power of the .38 RNL load was, particularly from a short barrel -- not just the date from yankeegunnuts, but the established fact that the .357 round was developed partly at the request of law enforcement, who were finding .38 revolvers inadequate against the cover provided by (gasp!) car bodies!
And as for the chronograph data: sure it's possible his machine was off. It's also possible it's perfectly accurate, and if people find that difficult to believe that's their problem. What's more likely, that a machine is wrong or that human perception is? You can decide, but a lot of what people think they know isn't so, and a lot of commonly held ideas are bunk. I've heard countless people tell me a .45 slug would "knock you on your ass" but if you know anything about physics, you'll know that's not so. It's entirely possible that the .38 slug could move at around 300fps, and the fact still not be all that noticeable because a .38 snubby is a belly gun, used the vast majority of time at bad breath distance against unarmored opponents, where even that low a velocity is adequate to inflict lethal wounds. After all, a Remington .41 rimfire Remington derringer would kill you across a card table, despite firing a 130 grain bullet at a measly 425fps. So why should it be any surprise that a 158gr .38 bullet moving somewhere between 300 and 500fps is deadly at similar ranges? And why should it also be any surprise that such a bullet would have a marked difficultly penetrating car bodies -- especially when this was a fact recording during the Depression era, and which led to the development of at least three cartridges designed specifically to address this shortcoming?