Download the data first.

In this data set, I already stored the residuals as  a list of lists, rub-res-vect
This can be obtained by
(def rub-res-vect (car (last (twoway-model-additive (1-many-list rub-y 7)))))

> (length rub-res-vect)
7
>

You also want to choose covariance (instead of correlation) when
the dialog box pops out.

> (def out-pca (pca-model  rub-res-vect))
 

==============================================================

        *** Principal Component Analysis ***

Scale type:                  "Covariances"

Number of observations:            12

Number of variables:               7

the first principal component:
(-0.5668170153682872 -0.37810281401058676 -0.18905618840697946 9.930992659915512E-5 0.18892955144582402 0.37798140585262596 0.5669657505608059)

the second principal component:
(-0.10428966501942284 0.4214173469891682 -0.5433042467497305 0.01096867333175929 0.5524411225206519 -0.4462647284185051 0.10903149734583985)

the third principal component:
(0.0025123144764022686 -0.0033646758525365636 0.4700382142333373 -0.3799731881991078 -0.08965373953012734 -0.5595244627008804 0.5599655375735848)

the companion output variances of PCA:
#(39.674751602626195 3.624364426767344E-4 1.2820043520744513E-5 8.240179638837601E-6 1.735057917979105E-6 8.279877116561998E-7 9.665699550076041E-30)

the proportions of all variances:
#(0.9999902694802282 9.135102334396778E-6 3.2312536959170195E-7 2.0769126773968922E-7 4.3731616831317E-8 2.0869183081428393E-8 2.4362107152186087E-31)

==============================================================
OUT-PCA
>

Notice that the second  eigenvalue is nearly zero.
This shows that the residuals are in fact one-dimensional. In other words, all rows are (nearly)
proportional to each other.
This tells us that there is a common trend in temperature for each level of chemical coupling agent used.