[BioC] differences between "manual" boxplots and "standard" boxplots of a oligo object

[Javier Pérez Florido]

Dear list,
I'm checking the quality of a set of raw Human Gene ST 1.0 arrays 
through the use of boxplots and histograms. CEL files are read using 
read.celfiles from oligo package.
Then, boxplots and histograms are plot using "boxplot" and "hist" 
commands respectively, but I observe some differences between the plots 
given by these commands and the plots of "manual" boxplots and 
histograms such these:

pms = log2(pm(OligoRaw))
dens = oligo:::matDensity(pms)
cols = darkColors(ncol(pms))
boxplot(as.data.frame(pms), range=0, col=cols)
matplot(dens[["x"]], dens[["y"]], type="l", lwd=2, 
col=cols,xlab="log-intensity", ylab="density")

where "OligoRaw" is the object that holds raw data from the CEL files.
To be more precise, in boxplots there are differences in the values 
related to Q3+1.5*IQR. In the "manual" boxplot, these values are bigger 
than in the standard boxplot command.
On the other hand, in histograms, the differences are related to the CEL 
files with the biggest density values. In the "manual" histogram 
(matplot), these values are bigger than in the standard hist command.

Why do these differences exist?
Thanks,
Javier