The following w.r.t. dino oils only: "The higher the VI, the more multigraded the oil" (How to read a can of oil (Part 1), by John S. Evans B.Sc., WearCheck Africa)/. Here are some oils (Chevron Supreme, Exxon Superflow, and Citgo Superguard) ordered by their VIs:
159 5w30 Chevron
158 5w30 Exxon
154 5w20 Citgo
154 5w30 Citgo
153 5w20 Exxon
148 5w20 Chevron
148 10w40 Chevron
148 10w40 Citgo
147 10w40 Exxon
137 10w30 Citgo
135 10w30 Chevron
135 10w30 Exxon
125 20w50 Exxon
122 20w50 Chevron
121 20w50 Citgo
The blank lines are there to emphasize the bigger drops in VI as you go down the table. It is clear that the 5w30 is the most multigraded of these various grades, with the 5w20s following a close second. The 10w40w follow close behind the 5w20s. Then there is a big gap to the much less multigraded 10w30 and another big gap from that to the 20w50. The very little data I have on 15w40 suggests it would follow close with the 10w30s or be somewhere between 10w30 and 20w50. If the viscosity index improver additive levels are roughly comparable to the VI (I know part of that is the natural VI of the base oil, but all else equal...), then 10w30 is the way to go except when extreme cold dictates a thinner oil. The idea of running 5w30 or 5w20 during hot summer weather appears absurd from this viewpoint. And the much maligned 10w40 ("too much viscosity spread," "suffers permanent viscosity collapse," etc.) appears actually to be more stable than the ever popular 5w oils.
Anyway, the quote from John Evans' article got me to thinking about this again. It seems a simple way to look at the viscosity index improver question. Even when you look at each of the three brands above separately (removes some, but not all, the unknowns) each separate brand still gives the same relationship I have described above.
Maybe we can add data on a few more brands and see how the order holds up. I purposefully avoided synthetic blends to minimize the unknown variable. Valvoline Durablend, for example, would skew the results because it has higher VIs due to its synthetic fraction.
[ February 27, 2004, 01:58 AM: Message edited by: TallPaul ]