Let $h$ be the unknown height of the tower. Let $d$ be the (equally unknown) distance from $P$, the point we first set up our equipment, to the base $B$ of the tower. Let $\theta$ be the measured angle of elevation. So in your case, $\theta$ is $26^\circ 50'$.
Draw the right-angled triangle $TBP$, where $T$ is the top of the tower. Then
Now repeat, letting $Q$ be the second point of measurement. The distance from $Q$ to the base of the tower is $d-25$. Let $\phi$ be the angle of elevation from $Q$. You were given $\phi$.
By the same reasoning as before, we have
We have two equations, in the two unknowns $h$ and $d$. We want to solve for $h$.
Take the reciprocal of each side of the first equation. We get
Similarly, take the reciprocal of each side of the second equation. We get
Look at the two equations $(1)$ and $(2)$. Subtract each side of $(2)$ from the corresponding side of $(1)$. We get
Now solving for $h$ is just some algebra. We get
Remark: You asked about attempting what I would call a linear approximation. Good idea. I will not go through the details of how one might try to do it that way. The problem is that in surveying, we try to achieve very high standards of accuracy. A rough estimate is not good enough. And we don't have to settle for a rough estimate, since, at least to the standards of prcision of our measurements, we can get an "exact" answer from Formula $(3)$.