Lemieux and McDavid are both much more dominant scorers than Howe was. Howe was a great scorer but he wasn't in the same league as Lemieux.
It's quite likely, or even certain (in my own view), that Lemieux was the more offensively gifted player. Lemieux certainly has three or four seasons (1988, 1989, 1993, probably 1996) that go beyond the offensive peak-ability of Gordie Howe, at least in as much as we can speculate with cross-era comparisons.
But "wasn't in the same League"...?? No, that's silly. The scoring environment in Howe's peak years --- say, autumn 1950 to spring 1955 or so (you could stretch it to 1957) --- was not only far lower than in Lemieux's best seasons, it was one of the lowest-such in NHL history. In Howe's (probably) very best season, 1952-53, he finished with 95 points, and the next non-Howe linemate in scoring was legendary Hall of Famer Rocket Richard with 61 points. Howe outscored the Rocket in 1953 by 45%, a ridiculous amount. I don't think there was any season where Lemieux outscored the next-best Canadian forward (non-linemate, or otherwise) by that much, even if you delete Gretzky and even if you project his 1993 totals to a full season.
During the four seasons 1950-51 to 1953-54, Howe outscored linemate Ted Lindsay 348 to 261 (by 25%), and again if you delete Lindsay then he outscored the next player (Richard) 348 to 238 (by
32%).
It's tricky to do this with with Lemieux because (a) Gretzky and (b) injuries, but if we pick 1987-88, 1988-89, 1989-90, 1991-92 and 1992-93 (which seems to me the fairest way to compare five seasons in a row or
almost in a row), and then ignore Gretky and isolate only Canadian competition, it comes out with Lemieux scoring 781 points in 317 games, or 2.46 PPG, or 984 points if he'd played the full five seasons. The next (non-Gretzky) player in production is Yzerman at 1.62 PPG, or 648 points if he'd played the full five seasons. Thus, Lemieux's points lead (projected) over Yzerman in five years would be 984 to 648 (by
52%).
So, as I would expect, Lemieux's (projected) lead over the next Canadian player is a large margin of 52% to 32%, considerably larger than Howe's.
However, as we all know, it's tough to compare with Howe over longer samples. I'm too lazy to run the numbers right now (having started this post seven hours ago and was taken away by family duties), but my guess is that if I did it over a ten-year or a twelve-year sample, the two players would be very close.
So, in conclusion:
-- when looking at a one-year peak, Howe's lead over non-linemate players is larger than Lemieux's.
-- when looking at best five years in a row (or almost in a row, in Mario's case), Lemieux's lead over non-Gretzky Canadian players is considerably larger than Howe's.
-- (I'm guessing that) when looking at a longer sample of 10 or 12 years, Howe's lead over non-linemate players is comparable to Mario's.
And of course, we're just looking at points here, which is only part of the story.