Unexpected behavior looping through sequence of dates

I understand that the two following loops should produce the exact same output. However, they do not. It appears that looping directly through the sequence of Date objects somehow makes them be coerced to numeric:

date1 = "20130301"
date2 = "20130302"

d1 = as.Date(date1, format="%Y%m%d", tz="GMT")
d2 = as.Date(date2, format="%Y%m%d", tz="GMT")

range = seq(from=d1, to=d2, by="day")
for (i in 1:length(range)) print(class(range[i]))

[1] "Date"
[1] "Date"

for (i in seq(from=d1, to=d2, by="day")) print(class(i))

[1] "numeric"
[1] "numeric"

This is output from R version 2.15.2 (2012-10-26) on Mac OS X 10.8.2.

Is this expected behavior?

It wasn't expected by me either, but it is what both my Mac R 2.15.3 and my Windows(32bit) 2.15.1 display. If you put those dates in a list, you do not get the coercion:

for (i in list(d1,d2)) print(class(i))

[1] "Date"
[1] "Date"

I was unable to find the reason for the original coercion in the help("for") page or the R Language Definition entry regarding for-loops. On the hunch that coercion via as.vector might be occurring, I tried with with a factor vector and did see that it was coerced to "character" which is the behavior that would be expected if that were happening behind the scenes:

 ff <- factor(letters[1:3])
 as.vector(ff)
#[1] "a" "b" "c"
 for (i in ff) print(class(i))
#[1] "character"
#[1] "character"
#[1] "character"

(I also tried searching for loop in the R Internals document and found nothing that explained this to me.)

David.
> -- 
> Alexandre Sieira
> CISA, CISSP, ISO 27001 Lead Auditor
> 
> "The truth is rarely pure and never simple."
> Oscar Wilde, The Importance of Being Earnest, 1895, Act I
> 
> 
> Sent with Sparrow (http://www.sparrowmailapp.com/?sig)
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

David Winsemius
Alameda, CA, USA