Surprising behavior of DataFrame.replace

This seems a bit surprising:

``` python
>>> from string import ascii_lowercase as letters
>>> from pandas import DataFrame
>>> df = DataFrame(list(letters[:4]), columns=['a'])
>>> df
   a
0  a
1  b
2  c
3  d
>>> df.replace({'a': 'b'})
   a
0  a
1  a
2  c
3  d
>>> df.replace({'a': 'c'})
   a
0  a
1  b
2  b
3  d
```

Does this have to do with padding?
