@@ -520,8 +520,7 @@ auto [S, T] = tf.parallel_for(
   v.end(),      // end of range
   [] (int i) { 
     std::cout << "parallel in " << i << '\n';
-  },
-  1  // execute one task at a time
+  }
 );
 // add dependencies via S and T.
 ```
@@ -545,6 +544,9 @@ auto [S, T] = tf.parallel_for(
 );
 ```
 
+By default, taskflow performs an even partition over worker threads
+if the group size is not specified.
+
 ### *reduce/transform_reduce*
 
 The method `reduce` creates a subgraph that applies a binary operator to a range of items.
 
@@ -32,7 +32,7 @@ void taskflow(int N) {
   tf::Taskflow tf;
   tf.parallel_for(range, [&] (const int i) { 
     printf("fib[%d]=%d\n", i, fib(i));
-  });
+  }, 1);
   tf.wait_for_all();
 
   auto tend = std::chrono::steady_clock::now();
 
@@ -6,7 +6,7 @@
 #include <taskflow/taskflow.hpp>  // the only include you need
 
 int main(){
-
+  
   tf::Taskflow tf(std::thread::hardware_concurrency());
 
   auto [A, B, C, D] = tf.silent_emplace(   //  the taskflow graph
 
@@ -625,10 +625,10 @@ class BasicFlowBuilder {
     auto silent_emplace(C&&...);
 
     template <typename I, typename C>
-    auto parallel_for(I, I, C&&, size_t = 1);
+    auto parallel_for(I, I, C&&, size_t = 0);
 
     template <typename T, typename C, std::enable_if_t<is_iterable_v<T>, void>* = nullptr>
-    auto parallel_for(T&, C&&, size_t = 1);
+    auto parallel_for(T&, C&&, size_t = 0);
 
     template <typename I, typename T, typename B>
     auto reduce(I, I, T&, B&&);
@@ -854,7 +854,7 @@ template <typename I, typename C>
 auto BasicFlowBuilder<NodeType>::parallel_for(I beg, I end, C&& c, size_t g) {
 
   using category = typename std::iterator_traits<I>::iterator_category;
-
+  
   if(g == 0) {
     auto d = std::distance(beg, end);
     auto w = std::max(size_t{1}, _num_workers);
@@ -1208,10 +1208,10 @@ class BasicTaskflow {
     auto silent_emplace(C&&...);
 
     template <typename I, typename C>
-    auto parallel_for(I, I, C&&, size_t = 1);
+    auto parallel_for(I, I, C&&, size_t = 0);
 
     template <typename T, typename C, std::enable_if_t<is_iterable_v<T>, void>* = nullptr>
-    auto parallel_for(T&, C&&, size_t = 1);
+    auto parallel_for(T&, C&&, size_t = 0);
 
     template <typename I, typename T, typename B>
     auto reduce(I, I, T&, B&&);