There are two differences that I am aware of between the August issue and the later issue. One of them is in the shading of the leaf under the Y of Ninety. Frankly I don't have good enough images to say anything definitive about that. But the easier difference to spot is that in the August issue the two diagonal lines at the top of the stamp have a gap between them of about a half a millimeter. It is a clear gap. (see http://www.theswedishtiger.com/62-scotts.html
) The later issue closed that gap and there appears to be a dot of ink right where the lines come together. Even in the images shown here it is clear there is no gap in either stamp and they both show the dot of ink. I have also read other papers on this topic and there does seem to be agreement that the 24, 30 and 90 cent stamps were all printed with the newer dies.
So the point is, just like the 24 cent and 30 cent stamps, both stamps on both pages were printed with the same die. There were subtle color differences but that is it.
So what seems wrong to me is the Scott numbering. They have chosen to use a single catalog number for the 24 cent and 30 cent stamps. 78P2 for both 24c stamps and 71P2 for both 30c stamps. But they have chosen to have two different catalog numbers for the 90c stamp. 72-E7c for the August page and 72P2 for the second page. They also show very different values, with the August stamp having double the value of the other.
All three stamps should be treated consistently in my mind. The 24, 30 and 90 cent stamps all have the same pattern. One die, and slight color differences. Personally I'd vote to have two numbers for each since they are different colors - perhaps two varieties - e.g. 78P2a and 78P2b. The strangest bit is the difference in value. I imagine these came from different sales, but if they are essentially the same stamp that price difference doesn't add up.