# Remove broken links in news.html

--- a/doc/news.html
+++ b/doc/news.html
@@ -286,8 +286,7 @@
     <ul>
     <li>Ability to swap the phase screen during a simulation run. This comes on top of the 
     "jump and reset" feature. In fact, it simply swaps (rotate) the screens every N resets 
-    (set this using loop.jumps2swapscreen, see the 
-    <a href="data-structures.html">data-structures</a> page).</li>
+    (set this using loop.jumps2swapscreen).</li>
     <li>At long last, I have figured out and implemented more realistic influence functions for
     stackarray (piezostack) deformable mirrors. One can set the coupling parameter of the
     influence functions. Acceptable values are from 4 to 30%.</li>
@@ -396,13 +395,13 @@
     <div id="blogentry">
     <div id="entrydate">4 April 2004</div>
     <div id="entrytitle">Update of the web site</div>
-    <div id="entrybody">I have updated the web site, adding new <a href="performance.html">performance results</a>, and the weblog/news (this very page).</div>
+    <div id="entrybody">I have updated the web site, adding new performance results, and the weblog/news (this very page).</div>
     </div>
 
     <div id="blogentry">
     <div id="entrydate">2004</div>
     <div id="entrytitle">Porting to FFTW!</div>
-    <div id="entrybody">Done ! I have relatively painleslly ported the fast (VE) C routines to using FFTW. It is almost as fast as the apple veclib (but not quite so, so I am keeping both). This means that YAO now runs on Linux! I have installed it on heze (2.8GHz Dual Athlon) and it gives nice results, although lono (Dual 2GHz G5) with the veclib FFTs is still faster (more details in <a href="performance.html">performance</a> page).</div>
+    <div id="entrybody">Done ! I have relatively painleslly ported the fast (VE) C routines to using FFTW. It is almost as fast as the apple veclib (but not quite so, so I am keeping both). This means that YAO now runs on Linux! I have installed it on heze (2.8GHz Dual Athlon) and it gives nice results, although lono (Dual 2GHz G5) with the veclib FFTs is still faster.</div>
     </div>
 
 
