Step 1: Use the "Mean minus Mode" form of the empirical relation instead of the "Mode equals" form.
The same well known empirical relation between mean, median and mode can be written in a different but equivalent shape, and starting from that shape changes the order of the arithmetic.
Step 2: Write down this form of the relation.
\[ \text{Mean} - \text{Mode} = 3(\text{Mean} - \text{Median}) \]
This says the gap between the mean and the mode is always three times the gap between the mean and the median, for a moderately skewed distribution.
Step 3: Compute the left-hand side using the given values.
We are given $\text{Mean} = 12$ and $\text{Mode} = 21$.
\[ \text{Mean} - \text{Mode} = 12 - 21 = -9 \]
Step 4: Solve for the gap between mean and median.
\[ -9 = 3(\text{Mean} - \text{Median}) \]
Divide both sides by 3:
\[ \text{Mean} - \text{Median} = -3 \]
Step 5: Solve for the median.
\[ \text{Median} = \text{Mean} - (-3) = \text{Mean} + 3 = 12 + 3 = 15 \]
Step 6: Final answer.
The median of the data is 15, which is option (C).
\[ \boxed{\text{Median} = 15} \]