diff --git a/README.Rmd b/README.Rmd
index a9c57b211..bbb12babb 100644
--- a/README.Rmd
+++ b/README.Rmd
@@ -74,9 +74,6 @@ str(vec_cast_common(FALSE, 1L, 2.5))
 The original motivation for vctrs comes from two separate but related problems. The first problem is that `base::c()` has rather undesirable behaviour when you mix different S3 vectors:
 
 ```{r}
-# combining factors makes integers
-c(factor("a"), factor("b"))
-
 # combining dates and date-times gives incorrect values; also, order matters
 dt <- as.Date("2020-01-01")
 dttm <- as.POSIXct(dt)
diff --git a/README.md b/README.md
index 7e85255cd..f5d1edf71 100644
--- a/README.md
+++ b/README.md
@@ -86,10 +86,6 @@ problems. The first problem is that `base::c()` has rather undesirable
 behaviour when you mix different S3 vectors:
 
 ``` r
-# combining factors makes integers
-c(factor("a"), factor("b"))
-#> [1] 1 1
-
 # combining dates and date-times gives incorrect values; also, order matters
 dt <- as.Date("2020-01-01")
 dttm <- as.POSIXct(dt)
diff --git a/vignettes/stability.Rmd b/vignettes/stability.Rmd
index d6cf480bb..71aa1db69 100644
--- a/vignettes/stability.Rmd
+++ b/vignettes/stability.Rmd
@@ -196,18 +196,17 @@ vec_c(Sys.Date(), factor("x"), "x")
 
 ### Factors
 
-Combining two factors returns an integer vector:
+Prior to 4.1.0 (May 2021), combining two factors returned an integer vector. The current behavior creates a new factor with a combination of factor levels, similar to `vec_c()`.
 
 ```{r}
 fa <- factor("a")
 fb <- factor("b")
 
 c(fa, fb)
+c(fb, fa)
 ```
 
-(This is documented in `c()` but is still undesirable.)
-
-`vec_c()` returns a factor taking the union of the levels. This behaviour is motivated by pragmatics: there are many places in base R that automatically convert character vectors to factors, so enforcing stricter behaviour would be unnecessarily onerous. (This is backed up by experience with `dplyr::bind_rows()`, which is stricter and is a common source of user difficulty.) 
+`vec_c()` also returns a factor taking the union of the levels. This behaviour is motivated by pragmatics: there are many places in base R that automatically convert character vectors to factors, so enforcing stricter behaviour would be unnecessarily onerous. (This is backed up by experience with `dplyr::bind_rows()`, which is stricter and is a common source of user difficulty.) 
 
 ```{r}
 vec_c(fa, fb)