JAL-1551 formatting

diff --git a/help/html/releases.html b/help/html/releases.html
index 485cf73..c331ae1 100755 (executable)
--- a/help/html/releases.html
+++ b/help/html/releases.html
@@ -228,7 +228,9 @@ li:before {
               <!-- JAL-2549 -->Updated JABAWS client to v2.2
             </li>
             <li>
-              <!-- JAL-2335 -->Filter non-standard amino acids and nucleotides when submitting to AACon and other MSA Analysis services
+              <!-- JAL-2335 -->Filter non-standard amino acids and
+              nucleotides when submitting to AACon and other MSA
+              Analysis services
             </li>
             <li>
               <!-- JAL-2316, -->URLs for viewing database
@@ -290,25 +292,23 @@ li:before {
               matrix - C->R should be '-3'<br />Old matrix restored
               with this one-line groovy script:<br />jalview.analysis.scoremodels.ScoreModels.instance.BLOSUM62.@matrix[4][1]=3
             </li>
-            <li><a name="2102scoremodelbugs"/>
-              <!-- JAL-2397 -->Fixed Jalview's treatment of gaps in PCA
-              and substitution matrix based Tree calculations.<br /> <br />In
-              earlier versions of Jalview, gaps matching gaps were
-              penalised, and gaps matching non-gaps penalised even more.
-              In the PCA calculation, gaps were actually treated as
-              non-gaps - so different costs were applied, which meant
-              Jalview's PCAs were different to those produced by
-              SeqSpace.<br />Jalview now treats gaps in the same way as
-              SeqSpace (ie it scores them as 0). <br /> <br />Enter
-              the following in the Groovy console to restore pre-2.10.2
-              behaviour:<br />
+            <li><a name="2102scoremodelbugs" /> <!-- JAL-2397 -->Fixed
+              Jalview's treatment of gaps in PCA and substitution matrix
+              based Tree calculations.<br /> <br />In earlier versions
+              of Jalview, gaps matching gaps were penalised, and gaps
+              matching non-gaps penalised even more. In the PCA
+              calculation, gaps were actually treated as non-gaps - so
+              different costs were applied, which meant Jalview's PCAs
+              were different to those produced by SeqSpace.<br />Jalview
+              now treats gaps in the same way as SeqSpace (ie it scores
+              them as 0). <br /> <br />Enter the following in the
+              Groovy console to restore pre-2.10.2 behaviour:<br />
               jalview.analysis.scoremodels.ScoreMatrix.scoreGapAsAny=true
               // for 2.10.1 mode <br />
               jalview.analysis.scoremodels.ScoreMatrix.scoreGapAsAny=false
               // to restore 2.10.2 mode <br /> <br /> <em>Note:
                 these settings will affect all subsequent tree and PCA
-                calculations (not recommended)</em>
-            </li>
+                calculations (not recommended)</em></li>
             <li>
               <!-- JAL-2424 -->Fixed off-by-one bug that affected
               scaling of branch lengths for trees computed using
@@ -661,7 +661,8 @@ li:before {
               doesn't always add secondary structure annotation.
             </li>
           </ul>
-        </div><tr>
+        </div>
+    <tr>
       <td width="60" nowrap>
         <div align="center">
           <strong><a name="Jalview.2.10.1">2.10.1</a><br /> <em>29/11/2016</em></strong>