Historical curiosity: why does var() allow as.character(<numeric>) input?

mean_var.R

# using a data.frame with a character column (not uncommon)
d <- data.frame(name = c("alpha", "beta", "gamma"),
                x    = runif(3),
                y    = runif(3))
d
#>    name         x         y
#> 1 alpha 0.8577527 0.4611686
#> 2  beta 0.2989644 0.9660795
#> 3 gamma 0.7221540 0.1730588

# trying to take the column-wise mean fails
apply(d, 2, mean)
#> Warning in mean.default(newX[, i], ...): argument is not numeric or
#> logical: returning NA

#> Warning in mean.default(newX[, i], ...): argument is not numeric or
#> logical: returning NA

#> Warning in mean.default(newX[, i], ...): argument is not numeric or
#> logical: returning NA
#> name    x    y 
#>   NA   NA   NA

# since apply converts (expectedly) to a matrix
# and thus coerces to most general type, which
# mean() chokes on

# var() HOWEVER:
apply(d, 2, var)
#> Warning in FUN(newX[, i], ...): NAs introduced by coercion
#>       name          x          y 
#>         NA 0.08495347 0.16113735

# only complains about the character column, 
# and otherwise coerces the numeric columns
# before applying the function

# using the numeric-only part of the data.frame works
apply(d[,-1], 2, mean)
#>         x         y 
#> 0.6262904 0.5334356
apply(d[,-1], 2, var)
#>          x          y 
#> 0.08495348 0.16113734

# Q: why does var() allow character input?
mean(as.character(runif(10)))
#> Warning in mean.default(as.character(runif(10))): argument is not numeric
#> or logical: returning NA
#> [1] NA
var(as.character(runif(10)))
#> [1] 0.04085411