diff --git a/README.md b/README.md
index b3d4e50..bbceb51 100644
--- a/README.md
+++ b/README.md
@@ -68,20 +68,18 @@ The disk format is stable as of 1.0.0.
 
 Future breaking changes will result in a major version bump and a migration path.
 
+## Run unit benchmarks
+
+```bash
+cargo bench --features bloom,lz4,miniz
+```
+
 ## License
 
 All source code is licensed under MIT OR Apache-2.0.
 
 All contributions are to be licensed as MIT OR Apache-2.0.
 
-## Development
-
-### Run benchmarks
-
-```bash
-cargo bench --features bloom --features lz4 --features miniz
-```
-
 ## Footnotes
 
 [1] https://rocksdb.org/blog/2017/05/12/partitioned-index-filter.html
diff --git a/src/memtable/mod.rs b/src/memtable/mod.rs
index b0abb51..07f9b20 100644
--- a/src/memtable/mod.rs
+++ b/src/memtable/mod.rs
@@ -71,16 +71,16 @@ impl Memtable {
         // NOTE: This range start deserves some explanation...
         // InternalKeys are multi-sorted by 2 categories: user_key and Reverse(seqno). (tombstone doesn't really matter)
         // We search for the lowest entry that is greater or equal the user's prefix key
-        // and has the highest seqno (because the seqno is stored in reverse order)
+        // and has the seqno (or lower) we want  (because the seqno is stored in reverse order)
         //
         // Example: We search for "abc"
         //
         // key -> seqno
         //
         // a   -> 7
-        // abc -> 5 <<< This is the lowest key (highest seqno) that matches the range
+        // abc -> 5 <<< This is the lowest key (highest seqno) that matches the key with seqno=None
         // abc -> 4
-        // abc -> 3
+        // abc -> 3 <<< If searching for abc and seqno=4, we would get this
         // abcdef -> 6
         // abcdef -> 5
         //