Popular Mechanics April 1934 Article, 10-W and 20-W "Winter" Oils

[SubieRubyRoo]

Unless you're buying Toyota OEM motor oil any "20 weight oil" you're buying from an auto parts store or Walmart nearly always has a viscosity over 9 cSt at 100c so your pretty much buying a thinner "30 weight oil".
The difference in fuel economy between a store bought 20 weight oil and a 30 weight won't even be detectable.
Most 20 weight oils I've looked at VOA tests for are usually 9 to 9.7 cSt and 30 weight oil VOA shows a viscosity of usually 10.5 cSt or less at 100c. Real world testing shows about a 1% difference in power and fuel economy.
Old timer wisdom and experience, indeed.

The difference between a 20 grade and a 40 grade isn’t detectable on a statistically significant level. There’s too many other variables in play that cannot be controlled. You’d likely need hundreds of thousands of miles on each oil to be able to say there was truly a difference.

 

[loneryder]

This is from the FSM for my '57 Plymouth...

View attachment 232584

Good old 10-30. Used in everything.

 

[oil pan 4]

The difference between a 20 grade and a 40 grade isn’t detectable on a statistically significant level. There’s too many other variables in play that cannot be controlled. You’d likely need hundreds of thousands of miles on each oil to be able to say there was truly a difference.

The cummins test I read showed a 3% difference but noted the valve train parts were wearing out about 30% faster using a 20 weight. Id take the 3% hit.

 

[SubieRubyRoo]

The cummins test I read showed a 3% difference but noted the valve train parts were wearing out about 30% faster using a 20 weight. Id take the 3% hit.

But is the Cummins recommended by the factory to run a 20 grade? Or was it done for a test, with the “normal” viscosity recommendation being a 30 grade??

 

[oil pan 4]

It was an old paper, probably from the early 2000s when 20 weights were all the rage for saving a tiny bit of fuel. The recommended viscosity was 30 and 40 weights depending "on read the manual".

 

[kschachn]

The difference between a 20 grade and a 40 grade isn’t detectable on a statistically significant level. There’s too many other variables in play that cannot be controlled. You’d likely need hundreds of thousands of miles on each oil to be able to say there was truly a difference.

The issue is with the inadequate method of measurement. A spectrographic analysis is subject to multiple and significant outside influences regardless of the length of the test. In fact, the longer length probably makes any determination less likely since over time the less significant influences become more significant. This is sometimes the case in testing.

We see that a lot on here when people say that you'd have to run that for a long time to be significant. However, there is no guarantee that would help. It's why a proper mathematical analysis is critical to any test.