Without a reproducible example it is hard to tell but it smells like the mother of all FAQs: the default display precision is four digits, so 16968 gets rounded to 16970.
Edit: We do need your sample data here because with a naive example I cannot reproduce this:
R> set.seed(42)
R> df <- data.frame(a=as.numeric(1:16968), b=16968:1,
+ c=rnorm(16968), d=runif(16968))
R> summary(df)
a b c d
Min. : 1 Min. : 1 Min. :-4.04328 Min. :0.000101
1st Qu.: 4243 1st Qu.: 4243 1st Qu.:-0.68271 1st Qu.:0.252515
Median : 8484 Median : 8484 Median :-0.00528 Median :0.505090
Mean : 8484 Mean : 8484 Mean :-0.00834 Mean :0.504563
3rd Qu.:12726 3rd Qu.:12726 3rd Qu.: 0.66746 3rd Qu.:0.758991
Max. :16968 Max. :16968 Max. : 4.32809 Max. :0.999976
Edit 2, with h/t to @SimonO101:
R> summary(df$a) ## what OP saw
Min. 1st Qu. Median Mean 3rd Qu. Max.
1 4240 8480 8480 12700 17000
R> summary(df$a, digits=6) ## what OP wanted to see
Min. 1st Qu. Median Mean 3rd Qu. Max.
1 4243 8484 8484 12726 16968
R>