On Dec 16, 5:37 pm, Peter Franks wrote:
[ . . . ]
Therefore, we are in the fourth year of a November cooling trend, correct?

In a word, "NO."

Let's analyze this biased world of your cherry picked data:

attach(aframe)
aframe

Year TempC
1 2005 15.15
2 2004 15.06
3 2006 15.03
4 2007 14.96

fitted.model - lm(TempC ~ Year)
summary(fitted.model)

Call:
lm(formula = TempC ~ Year)
Residuals:
1 2 3 4
0.079 -0.053 0.001 -0.027

Coefficients:
Estimate Std. Error t value Pr(|t|)
(Intercept) 99.281000 62.717994 1.58297 0.25426
Year -0.042000 0.031273 -1.34301 0.31138

Residual standard error: 0.0699285 on 2 degrees of freedom
Multiple R-Squared: 0.474194, Adjusted R-squared: 0.21129
F-statistic: 1.80368 on 1 and 2 DF, p-value: 0.311383

You're only 31% sure that this trendline explains 21%
of the variance. GIVE IS A BREAK, and forget your
imaginary cooling trends, OK Peter!