diff --git a/README.md b/README.md
index 23b98af..21164da 100644
--- a/README.md
+++ b/README.md
@@ -156,16 +156,18 @@ of the `aquatic_http` section above.
 #### Benchmarks
 
 Performance was compared to
-[wt-tracker](https://github.com/Novage/wt-tracker) and [bittorrent-tracker](https://github.com/webtorrent/bittorrent-tracker), using
-`aquatic_ws_load_test` with one worker, 16 connections and 2 offers per request. Both comparison trackers reached 100% single-core utilization.
+[wt-tracker](https://github.com/Novage/wt-tracker) and [bittorrent-tracker](https://github.com/webtorrent/bittorrent-tracker) using
+`aquatic_ws_load_test` with one worker, 16 connections and 2 offers per request.
 
-Server responses per second, best result in bold:
+| tracker            | responses per second |
+| ------------------ | -------------------- |
+| aquatic            |  __126k__            |
+| wt-tracker         |  54k                 |
+| bittorrent-tracker |  28k                 |
 
-|   aquatic   |  wt-tracker  | bittorrent-tracker |
-| ----------- | ------------ | ------------------ |
-|  __126k__   |     54k      |        28k         |
+Both reference trackers reached 100% single-core utilization.
 
-This benchmark is not very realistic, as it simulates a tiny number of clients each sending a large number of requests, but I still think that it gives a useful indication of relative performance.
+This benchmark is not very realistic, as it simulates a tiny number of clients, each sending a large number of requests. Nontheless, I think that it gives a useful indication of relative performance.
 
 ## Load testing