diff --git a/code/numerical/numerical.c b/code/numerical/numerical.c
index 3c1e5ca9..4f43f85d 100644
--- a/code/numerical/numerical.c
+++ b/code/numerical/numerical.c
@@ -68,16 +68,10 @@ static mp_obj_t numerical_sum_mean_std_iterable(mp_obj_t oin, uint8_t optype, si
     size_t count = 0;
     mp_obj_iter_buf_t iter_buf;
     mp_obj_t item, iterable = mp_getiter(oin, &iter_buf);
-    if((item = mp_iternext(iterable)) != MP_OBJ_STOP_ITERATION) {
-        value = mp_obj_get_float(item);
-        sum += value;
-        M = m = value;
-        count++;
-    }
     while((item = mp_iternext(iterable)) != MP_OBJ_STOP_ITERATION) {
         value = mp_obj_get_float(item);
         sum += value;
-        m = M + (value - M) / count;
+        m = M + (value - M) / (count + 1);
         s = S + (value - M) * (value - m);
         M = m;
         S = s;
@@ -88,7 +82,7 @@ static mp_obj_t numerical_sum_mean_std_iterable(mp_obj_t oin, uint8_t optype, si
     } else if(optype == NUMERICAL_MEAN) {
         return count > 0 ? mp_obj_new_float(m) : mp_obj_new_float(MICROPY_FLOAT_CONST(0.0));
     } else { // this should be the case of the standard deviation
-        return count > ddof ? mp_obj_new_float(MICROPY_FLOAT_C_FUN(sqrt)(count * s / (count - ddof))) : mp_obj_new_float(MICROPY_FLOAT_CONST(0.0));
+        return count > ddof ? mp_obj_new_float(MICROPY_FLOAT_C_FUN(sqrt)(s / (count - ddof))) : mp_obj_new_float(MICROPY_FLOAT_CONST(0.0));
     }
 }
 
diff --git a/code/ulab.c b/code/ulab.c
index eec70c1f..2a34eb6d 100644
--- a/code/ulab.c
+++ b/code/ulab.c
@@ -32,7 +32,7 @@
 #include "user/user.h"
 #include "vector/vectorise.h"
 
-#define ULAB_VERSION 1.7.1
+#define ULAB_VERSION 1.7.2
 #define xstr(s) str(s)
 #define str(s) #s
 #if ULAB_NUMPY_COMPATIBILITY
diff --git a/code/ulab.h b/code/ulab.h
index 7f8fdfe0..b64bfbd4 100644
--- a/code/ulab.h
+++ b/code/ulab.h
@@ -74,8 +74,8 @@
 // the ndarray binary operators
 #define NDARRAY_HAS_BINARY_OPS              (1)
 
-// Firmware size can be reduced at the expense of speed by using function 
-// pointers in iterations. For each operator, he function pointer saves around 
+// Firmware size can be reduced at the expense of speed by using function
+// pointers in iterations. For each operator, he function pointer saves around
 // 2 kB in the two-dimensional case, and around 4 kB in the four-dimensional case.
 
 #define NDARRAY_BINARY_USES_FUN_POINTER     (0)
@@ -209,7 +209,7 @@
 #ifndef ULAB_VECTORISE_MODULE
 #define ULAB_VECTORISE_MODULE           (1)
 
-// Firmware size can be reduced at the expense of speed by using a function 
+// Firmware size can be reduced at the expense of speed by using a function
 // pointer in iterations. Setting ULAB_VECTORISE_USES_FUNCPOINTER to 1 saves
 // around 800 bytes in the four-dimensional case, and around 200 in two dimensions.
 #define ULAB_VECTORISE_USES_FUN_POINTER (1)
diff --git a/docs/ulab-change-log.md b/docs/ulab-change-log.md
index 8589de56..1e71feeb 100644
--- a/docs/ulab-change-log.md
+++ b/docs/ulab-change-log.md
@@ -1,3 +1,9 @@
+Fri, 29 Jan 2021
+
+version 1.7.2
+
+    fixed code calculating the standard deviation of iterables
+
 Fri, 15 Jan 2021
 
 version 1.7.1