diff --git a/src/metrics/histogram.rs b/src/metrics/histogram.rs
index 177b49a..e17799c 100644
--- a/src/metrics/histogram.rs
+++ b/src/metrics/histogram.rs
@@ -14,6 +14,20 @@ use std::sync::{Arc, Mutex, MutexGuard};
 /// let histogram = Histogram::new(exponential_buckets(1.0, 2.0, 10));
 /// histogram.observe(4.2);
 /// ```
+///
+/// [`Histogram`] does not implement [`Default`], given that the choice of
+/// bucket values depends on the situation [`Histogram`] is used in. As an
+/// example, to measure HTTP request latency, the values suggested in the
+/// Golang implementation might work for you:
+///
+/// ```
+/// # use prometheus_client::metrics::histogram::Histogram;
+/// // Default values from go client(https://github.com/prometheus/client_golang/blob/5d584e2717ef525673736d72cd1d12e304f243d7/prometheus/histogram.go#L68)
+/// let histogram = Histogram::new(IntoIterator::into_iter([
+///    0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1.0, 2.5, 5.0, 10.0,
+/// ]));
+/// histogram.observe(4.2);
+/// ```
 // TODO: Consider using atomics. See
 // https://github.com/tikv/rust-prometheus/pull/314.
 pub struct Histogram {