diff --git a/hadoop-tools/hadoop-aws/src/test/java/org/apache/hadoop/fs/s3a/select/ITestS3SelectLandsat.java b/hadoop-tools/hadoop-aws/src/test/java/org/apache/hadoop/fs/s3a/select/ITestS3SelectLandsat.java
index 8b0578d..51da971 100644
--- a/hadoop-tools/hadoop-aws/src/test/java/org/apache/hadoop/fs/s3a/select/ITestS3SelectLandsat.java
+++ b/hadoop-tools/hadoop-aws/src/test/java/org/apache/hadoop/fs/s3a/select/ITestS3SelectLandsat.java
@@ -415,7 +415,7 @@
       long increment = 64 * _1KB;
 
       // seek forward, comparing bytes
-      for(offset = 32 * _1KB; offset < actualLen; offset += increment) {
+      for(offset = 32 * _1KB; offset < _1MB; offset += increment) {
         seek(seekStream, offset);
         assertEquals("Seek position in " + seekStream,
             offset, seekStream.getPos());
@@ -423,11 +423,6 @@
         assertEquals("byte at seek position",
             dataset[(int) seekStream.getPos()], seekStream.read());
       }
-      for(; offset < len; offset += _1MB) {
-        seek(seekStream, offset);
-        assertEquals("Seek position in " + seekStream,
-            offset, seekStream.getPos());
-      }
       // there's no knowledge of how much data is left, but with Gzip
       // involved there can be a lot. To keep the test duration down,
       // this test, unlike the simpler one, doesn't try to read past the