diff --git a/README.md b/README.md
index 4e1517be..5593ce3b 100644
--- a/README.md
+++ b/README.md
@@ -29,9 +29,9 @@ addresses is done at your own risk.
 > recovered from the mnemonic.
 
 > The wallet and seed are short, simple text files suitable for printing
-> or even writing out by hand.  Their built-in checksums can be used to
-> verify they've been correctly copied.  The base-58-encoded seed is
-> short enough to memorize, providing another brain storage alternative.
+> or even writing out by hand.  Built-in checksums are used to verify
+> they've been correctly copied.  The base-58-encoded seed is short
+> enough to memorize, providing another brain storage alternative.
 
 > Implemented as a suite of lightweight python scripts with a
 > command-line interface, MMGen demands practically no system resources.
@@ -50,13 +50,11 @@ addresses is done at your own risk.
 >  Install mmgen:
 
             git clone https://github.com/mmgen/mmgen.git
-
             cd mmgen; sudo ./setup.py install
 
 >  Install vanitygen (optional but recommended):
 
             git clone https://github.com/samr7/vanitygen.git
-
             (build and put the 'keyconv' executable in your path)
 
 ### Getting Started:
@@ -138,7 +136,6 @@ addresses is done at your own risk.
 	        $ git clone https://github.com/sipa/bitcoin
             $ git branch mywatchonly remotes/origin/watchonly
             $ git checkout mywatchonly
-
             (build, install)
 
 > With your newly-compiled bitcoind running, import the addresses from
@@ -231,10 +228,10 @@ addresses is done at your own risk.
 
 ### Test suite:
 > To see what tests are available, run the scripts in the 'tests'
-> directory without arguments.  Some may find the following tests
+> directory with no arguments.  Some may find the following tests
 > interesting:
 
->> Compare 10 addresses generated with 'keyconv' with internally-generated ones:
+>> Compare 10 addresses generated by 'keyconv' with internally-generated ones:
 >>> `tests/bitcoin.py keyconv_compare_randloop 10`
 
 >> Perform 1000 hex -> base58 -> hex conversions, comparing results: