With the default scale, you see that numbers left of the bar go up by two - hence anything after the 4 is forty-something or fifty-something:
> stem(d,scale=1)
The decimal point is 1 digit(s) to the right of the |
0 | 577
2 | 7
4 | 19
6 | 0325
8 | 50378
Using scale=2
, you'll see numbers left of the bar go up by one, so now you can get exact reconstruction of your input, since you input integers:
> stem(d,scale=2)
The decimal point is 1 digit(s) to the right of the |
0 | 57
1 | 7
2 |
3 | 7
4 | 1
5 | 9
6 | 03
7 | 25
8 | 5
9 | 0378
Going further you can even split it by first and second five within each decade:
> stem(d,scale=4)
The decimal point is 1 digit(s) to the right of the |
0 | 57
1 |
1 | 7
2 |
2 |
3 |
3 | 7
4 | 1
4 |
5 |
5 | 9
6 | 03
6 |
7 | 2
7 | 5
8 |
8 | 5
9 | 03
9 | 78
Stem plots do not always guarantee you can reproduce the data by reversing the process, and that's not what they're for.