Compute ln 3 approximately, then use the result
Compute $\ln 3$ approximately, then use the result to compute $3^{\sqrt{4.3}}$.
Explanation of Method
Three linearisations chained together, and how their errors interact.
Step 1: Linearise the logarithm at $x = e$
The nearest point at which $\ln x$ is known exactly is $x = e$, where $\ln e = 1$ and $\bigl(\ln x\bigr)' = \dfrac1x$ equals $\dfrac1e$: \[\ln x \approx 1+\frac{x-e}{e} = \frac{x}{e}. \tag{1}\]
The linearisation of the logarithm at $e$ is therefore just division by $e$ — a memorable form.
Step 2: Evaluate at $x = 3$
\[ \ln 3 \approx \frac{3}{e} = \frac{3}{2.718282} \approx 1.10364. \tag{2} \]
The true value is $1.09861$, so this is high by $0.00503$ — about $0.46\%$. The error is respectable given that $3$ is not especially close to $e$.
Step 3: The exponent, from Problem 260
\[ \sqrt{4.3} = \sqrt{2^{2}+0.3} \approx 2+\frac{0.3}{4} = 2.075. \tag{3} \]
Step 4: Split off the integer part
Writing $2.075 = 2+0.075$ isolates a small exponent: \[3^{2.075} = 3^{2}\cdot3^{0.075} = 9\cdot3^{0.075}. \tag{4}\]
Step 5: Linearise $3^{\alpha}$ at $\alpha = 0$
With $g(\alpha) = 3^{\alpha} = e^{\alpha\ln3}$ we get $g(0) = 1$ and $g'(0) = \ln3$, so
\[ 3^{\alpha} \approx 1+\alpha\ln3, \qquad 3^{0.075} \approx 1+0.075\times1.10364 = 1.08277. \tag{5} \]
Step 6: Combine
\[ 3^{\sqrt{4.3}} \approx 9\times1.0828 = 9.745. \]
Step 7: Verification
\[\text{Exact: } \sqrt{4.3} = 2.0736441,\ \ln3 = 1.0986123,\ \text{product } = 2.2781309\]
\[3^{\sqrt{4.3}} = e^{2.2781309} = 9.7584242\]
\[\text{Estimate } 9.745 \text{ is low by } 0.0132, \text{ i.e. } 0.135\% \;\;\checkmark\]
Result
Final answer:
\[\boxed{\,\ln 3 \approx 1.104, \qquad 3^{\sqrt{4.3}} \approx 9.745\,}\]
Remark
Remark. Three separate approximations are chained here, and it is worth noting that their errors do not simply accumulate. Steps (2) and (3) both overestimate — $\ln3$ by $0.5\%$ and $\sqrt{4.3}$ by $0.07\%$ — while step (5) underestimates, since $3^{0.075} = 1.08589$ against the linear $1.08277$. The final $0.135\%$ error is smaller than the error in $\ln 3$ alone, because the two effects run in opposite directions.