Harmonic Series

The central idea

The harmonic series $\sum_{n=1}^{\infty}1/n$ diverges, even though its terms approach zero. More generally, the $p$-series $\sum1/n^p$ converges exactly when $p>1$ and diverges when $p\le1$. The harmonic case $p=1$ is the threshold.

Definitions, hypotheses, and notation

The integral $\int_1^N dx/x=\ln N$ gives another proof and describes the slow growth rate. Harmonic partial sums are approximately $\ln N+\gamma$, so even a very long numerical computation can misleadingly appear stable. Convergence is an infinite-tail property, not a judgment from small consecutive differences.

Removing or adding finitely many terms cannot change divergence. Some sparse subseries can converge, but retaining a fixed positive fraction of harmonic-scale terms generally preserves divergence. The standard harmonic series is therefore a central benchmark for comparison tests.

Conceptual meaning

The terms shrink too slowly for their accumulated total to settle. Grouping successive powers-of-two blocks reveals that each block contributes at least $1/2$, so partial sums eventually exceed every bound.

A dependable method and decision rule

1. Recognize exact or comparable $1/n$ behavior.
2. Use the $p$-series criterion when the exponent is explicit.
3. For nearby rational forms, choose direct or limit comparison with $1/n$.
4. Do not let the zero term limit suggest convergence.
5. State divergence; a divergent series has no finite sum.

Fully worked example

Graphical or geometric meaning

A graph of partial sums rises extremely slowly, which can look nearly flat in a finite window. The grouping proof zooms out by blocks and shows a repeated minimum vertical gain that no finite plot can erase.

Common mistakes and why they fail

Verification and reasonableness checks

• Compare the term pattern with the $p=1$ threshold.
• Calculate block lower bounds rather than relying on a plot.
• Distinguish the sequence $1/n$, which converges to zero, from its series.

Small terms can still accumulate without bound

Although $1/n\to0$, the harmonic sum diverges. Group terms from $2^k+1$ through $2^{k+1}$: the block has $2^k$ terms, each at least $1/2^{k+1}$, so every block contributes at least $1/2$. Infinitely many blocks force unbounded partial sums. This anchors the $p$-series rule that $\sum1/n^p$ converges exactly for $p>1$. When comparing a new positive series to a $p$-series, preserve inequality direction and constants carefully. Logarithmic modifications near $p=1$ often require an integral or condensation argument rather than an informal claim that the terms “look harmonic.” The term test alone remains inconclusive here.

Practice

1. Does $\sum1/n^{1.1}$ converge?
2. Does $\sum1/\sqrt n$ converge?
3. Does starting the harmonic series at $n=100$ change divergence?

Connections and next steps

• Divergence Test
• Integral Test
• Comparison Test