The most recent version of lme4 (not the one I've been using from 2014) objects to two-wave data.
I confirmed this by starting a new fully updated R-version with a newly downloaded lme4,
which for the Brogan-Kutner example
> bk = read.table(file="http://statweb.stanford.edu/~rag/stat222/brogkutlong2.dat", header = T)
> bk$t1 = bk$time - 1
> bklmera = lmer(outcome ~ t1 + t1:as.factor(method) + (t1|subject), data = bk)
Error: number of observations (=42) <= number of random effects (=42) for term (t1 | subject); the random-effects parameters and the residual variance (or scale parameter) are probably unidentifiable
A help thread that indicated appending control = lmerControl(check.nobs.vs.nRE = "warning") in the lmer statement will get you an functional lmer object that you can do summary on and get fixed effects. Random effects and CI for such appear not to work well.
https://github.com/lme4/lme4/issues/175
The work-arounds are to use the control statement for lmer OR to employ the older brother of lme4, package nlme, function lme for two-wave data. The nlme package is part of base R and is still widely used (in fact the brand new book 'Multilevel models with R' annoyingly uses nlme as the primary).
The code above changes to (notice the clunkier syntax for the random part of the mixed-model).
> bklmea = lme(outcome ~ t1 + t1:as.factor(method), random = ~ t1|subject, data = bk)
> summary(bklmea)
A short session using lmer and lme for the Brogan-Kutner data is provided here