diff --git a/pandas/core/computation/expressions.py b/pandas/core/computation/expressions.py
index 68d08fb476950..0e9077e6d557e 100644
--- a/pandas/core/computation/expressions.py
+++ b/pandas/core/computation/expressions.py
@@ -223,7 +223,7 @@ def evaluate(op, a, b, use_numexpr: bool = True):
     use_numexpr : bool, default True
         Whether to try to use numexpr.
     """
-    op_str = _op_str_mapping.get(op, None)
+    op_str = _op_str_mapping[op]
     if op_str is not None:
         use_numexpr = use_numexpr and _bool_arith_check(op_str, a, b)
         if use_numexpr:
diff --git a/pandas/core/ops/__init__.py b/pandas/core/ops/__init__.py
index 3ca160e39787e..18a3d93bd5c3b 100644
--- a/pandas/core/ops/__init__.py
+++ b/pandas/core/ops/__init__.py
@@ -280,9 +280,8 @@ def dispatch_to_series(left, right, func, axis=None):
         bm = left._mgr.operate_blockwise(right._mgr, array_op)
         return type(left)(bm)
 
-    elif isinstance(right, ABCSeries) and axis == "columns":
-        # We only get here if called via _combine_series_frame,
-        # in which case we specifically want to operate row-by-row
+    elif isinstance(right, ABCSeries) and axis == 1:
+        # axis=1 means we want to operate row-by-row
         assert right.index.equals(left.columns)
 
         if right.dtype == "timedelta64[ns]":
@@ -292,6 +291,8 @@ def dispatch_to_series(left, right, func, axis=None):
             right = np.asarray(right)
         else:
             right = right._values
+            # maybe_align_as_frame ensures we do not have an ndarray here
+            assert not isinstance(right, np.ndarray)
 
         arrays = [array_op(l, r) for l, r in zip(left._iter_column_arrays(), right)]
 
@@ -440,35 +441,6 @@ def flex_wrapper(self, other, level=None, fill_value=None, axis=0):
 # DataFrame
 
 
-def _combine_series_frame(left, right, func, axis: int):
-    """
-    Apply binary operator `func` to self, other using alignment and fill
-    conventions determined by the axis argument.
-
-    Parameters
-    ----------
-    left : DataFrame
-    right : Series
-    func : binary operator
-    axis : {0, 1}
-
-    Returns
-    -------
-    result : DataFrame or Dict[int, Series[]]
-    """
-    # We assume that self.align(other, ...) has already been called
-
-    rvalues = right._values
-    assert not isinstance(rvalues, np.ndarray)  # handled by align_series_as_frame
-
-    if axis == 0:
-        new_data = dispatch_to_series(left, right, func)
-    else:
-        new_data = dispatch_to_series(left, right, func, axis="columns")
-
-    return new_data
-
-
 def _align_method_FRAME(
     left, right, axis, flex: Optional[bool] = False, level: Level = None
 ):
@@ -671,7 +643,7 @@ def f(self, other, axis=default_axis, level=None, fill_value=None):
 
         elif isinstance(other, ABCSeries):
             axis = self._get_axis_number(axis) if axis is not None else 1
-            new_data = _combine_series_frame(self, other, op, axis=axis)
+            new_data = dispatch_to_series(self, other, op, axis=axis)
         else:
             # in this case we always have `np.ndim(other) == 0`
             if fill_value is not None:
@@ -707,7 +679,7 @@ def f(self, other, axis=default_axis, level=None):
 
         elif isinstance(other, ABCSeries):
             axis = self._get_axis_number(axis) if axis is not None else 1
-            new_data = _combine_series_frame(self, other, op, axis=axis)
+            new_data = dispatch_to_series(self, other, op, axis=axis)
         else:
             # in this case we always have `np.ndim(other) == 0`
             new_data = dispatch_to_series(self, other, op)
@@ -730,7 +702,7 @@ def f(self, other):
             self, other, axis=None, level=None, flex=False
         )
 
-        axis = "columns"  # only relevant for Series other case
+        axis = 1  # only relevant for Series other case
         # See GH#4537 for discussion of scalar op behavior
         new_data = dispatch_to_series(self, other, op, axis=axis)
         return self._construct_result(new_data)
diff --git a/pandas/core/ops/array_ops.py b/pandas/core/ops/array_ops.py
index cb1b2f0c37c6f..3379ee56b6ad0 100644
--- a/pandas/core/ops/array_ops.py
+++ b/pandas/core/ops/array_ops.py
@@ -85,13 +85,7 @@ def masked_arith_op(x: np.ndarray, y, op):
         yrav = y.ravel()
         mask = notna(xrav) & ymask.ravel()
 
-        if yrav.shape != mask.shape:
-            # FIXME: GH#5284, GH#5035, GH#19448
-            # Without specifically raising here we get mismatched
-            # errors in Py3 (TypeError) vs Py2 (ValueError)
-            # Note: Only = an issue in DataFrame case
-            raise ValueError("Cannot broadcast operands together.")
-
+        # See GH#5284, GH#5035, GH#19448 for historical reference
         if mask.any():
             with np.errstate(all="ignore"):
                 result[mask] = op(xrav[mask], yrav[mask])
@@ -378,13 +372,28 @@ def get_array_op(op):
         # TODO: avoid getting here
         return op
 
-    op_name = op.__name__.strip("_")
+    op_name = op.__name__.strip("_").lstrip("r")
+    if op_name == "arith_op":
+        # Reached via DataFrame._combine_frame
+        return op
+
     if op_name in {"eq", "ne", "lt", "le", "gt", "ge"}:
         return partial(comparison_op, op=op)
     elif op_name in {"and", "or", "xor", "rand", "ror", "rxor"}:
         return partial(logical_op, op=op)
-    else:
+    elif op_name in {
+        "add",
+        "sub",
+        "mul",
+        "truediv",
+        "floordiv",
+        "mod",
+        "divmod",
+        "pow",
+    }:
         return partial(arithmetic_op, op=op)
+    else:
+        raise NotImplementedError(op_name)
 
 
 def maybe_upcast_datetimelike_array(obj: ArrayLike) -> ArrayLike: