diff --git a/Fasterlabeling.md b/Fasterlabeling.md
index 822ff2c..d0fda92 100644
--- a/Fasterlabeling.md
+++ b/Fasterlabeling.md
@@ -19,4 +19,4 @@ avg: 1.5799ms | min: 1.6617ms | total: 16.766ms <-- blank png8:z=1:m=h
 
 * `label cache` - are your labels coming from multiple layers and being rendered separately from other data? This is a common pattern with Mapnik stylesheets and is usually a good strategy because then layer queries for labels can be customized (e.g. only might pull the `name` column, or you might restrict using a where clause like `WHERE name is not null`). But if your stylesheet has many layers and some of the other layers hit the same tables and rows that your labels come from (for line drawing) then you can reduce the i/o burden on the database by moving your `TextSymbolizer` to the same `Rule` as your `LineSymbolizer`. Then the problem becomes that some labels with get rendered underneath some lines. This is where label caching could help, see: [#870](https://github.com/mapnik/mapnik/issues/870).
 
-* Want sparse labels? Only rendering a few labels here and there is often stylistically desirable over rendering dense labels. It is also potentially very helpful for overall labeling performance, because usually (ideally) label drawing is more expensive than pulling label data (i/o) and doing label placement logic, so less drawing should equal measurably faster maps overall. But, good stylesheet design is needed to ensure this assumption holds: that i/o and placement logic are not the bottlenecks. In terms of placement logic, if you know you want sparse labels on lines, then setting a very high `max-char-angle-delta` can help knock out labels early on from being placed (the default is `22.5`). Also setting an aggressive `minimum-path-length` ([#865](https://github.com/mapnik/mapnik/issues/)) can also help avoid Mapnik ever attempting to process short lines. Setting `minimum-distance` can also help, but this check comes at the end of most other placement logic, so while it might still help avoid labels being rendering it is not much of a fast track for avoiding placement logic overhead. Also, if you do not set the `spacing` parameter then labeling long lines should not be much more expensive than other lengths, but beware that small spacing values (like 1-50) will likely result in placement logic on lines greater than 50 pixels long becoming a bottleneck.
\ No newline at end of file
+* Want sparse labels? Only rendering a few labels here and there is often stylistically desirable over rendering dense labels. It is also potentially very helpful for overall labeling performance, because usually (ideally) label drawing is more expensive than pulling label data (i/o) and doing label placement logic, so less drawing should equal measurably faster maps overall. But, good stylesheet design is needed to ensure this assumption holds: that i/o and placement logic are not the bottlenecks. In terms of placement logic, if you know you want sparse labels on lines, then setting a very high `max-char-angle-delta` can help knock out labels early on from being placed (the default is `22.5`). Also setting an aggressive `minimum-path-length` ([#865](https://github.com/mapnik/mapnik/issues/865)) can also help avoid Mapnik ever attempting to process short lines. Setting `minimum-distance` can also help (but make sure [#995](https://github.com/mapnik/mapnik/issues/995) is fixed in your version of mapnik), however this check comes at the end of most other placement logic, so while it might still help avoid labels being rendering it is not much of a fast track for avoiding placement logic overhead. Also, if you do not set the `spacing` parameter then labeling long lines should not be much more expensive than other lengths, but beware that small spacing values (like 1-50) will likely result in placement logic on lines greater than 50 pixels long becoming a bottleneck.
\ No newline at end of file