diff --git a/GPy/likelihoods/likelihood_functions.py b/GPy/likelihoods/likelihood_functions.py
index 81d93f6b..25f770b5 100644
--- a/GPy/likelihoods/likelihood_functions.py
+++ b/GPy/likelihoods/likelihood_functions.py
@@ -560,7 +560,7 @@ class gaussian(likelihood_function):
         """
         assert y.shape == f.shape
         e = y - f
-        dlik_dsigma = -0.5*self.D/self._variance - 0.5*np.trace(np.dot(e.T, np.dot(self.I, e)))
+        dlik_dsigma = -0.5*self.N/self._variance - 0.5*np.trace(np.dot(e.T, np.dot(self.I, e)))
         return dlik_dsigma
 
     def dlik_df_dstd(self, y, f, extra_data=None):
@@ -579,7 +579,7 @@ class gaussian(likelihood_function):
         $$\frac{d}{d\sigma}(\frac{d^{2}p(y_{i}|f_{i})}{d^{2}f}) = \frac{2\sigma v(v + 1)(\sigma^2 v - 3(y-f)^2)}{((y-f)^2 + \sigma^2 v)^3}$$
         """
         assert y.shape == f.shape
-        dlik_hess_dsigma = np.diag(1.0/(self._variance**2)*self.I)[:, None]
+        dlik_hess_dsigma = np.diag((1.0/(self._variance**2))*self.I)[:, None]
         return dlik_hess_dsigma
 
     def _gradients(self, y, f, extra_data=None):